Bulletin STATE OF ILLINOIS Adlai E. Stevenson, Governor DEPARTMENT OF REGISTRATION AND EDUCATION Noble J. Puffer, Director NATURAL HISTORY SURVEY DIVISION Harlow B. Mills, Chief Volume 25 B U L L E T I N Article 3 Canada Geese of the Mississippi Flyway tVith Special Reference to an Illinois Flock HAROLD C. HANSON ROBERT H. SMITH Printed hy .lulhorily of the State of Illinois URBANA, ILLINOIS March 1950 STATE OF ILLINOIS Adlai E. Stevenson, Governor DEPARTMENT OF REGISTRATION AND EDUCATION Noble J. Puffer, Director BOARD OF NATURAL RESOURCES AND CONSERVATION A. E. Emerson, Ph.D., Biology L. H. Tiffany, Ph.D., Forestry L R. HowsoN, B.S.C.E., C.E., Engineering Noble J. Puffer, Chairman George D. Stoddard, Ph.D., Litt.D., L.H.D., LL.D., President of the University of Illinois Walter H. Newhouse, Ph.D., Geology Roger Adams, Ph.D., D.Sc, Chemistry NATURAL HISTORY SURVEY DIVISION Urbana, Illinois Scientific and Technical Staff Harlow B. Mills, Ph.D., Chief Bessie B. Henderson, M.S., Assistant to the Chief Section of Economic Entomology George C. Decker, Ph.D., Entomologist and Head J. H. Bigger, M.S., Entomologist L. L. English, Ph.D., Entomologist C. J. Weinman, Ph.D., Entomologist S. C. Chandler, B.S., Associate Entomologist Willis N. Bruce, M.A., Assistant Entomologist John M. Wright, M.A., Assistant Entomologist H. B. Petty, M.A., Associate in Entomology Extension James M. Bann, B.S., Research Assistant Section of Faunistlc Surveys and Insect Identification H. H. Ross, Ph.D., Systematic Entomologist and Head Milton W. Sanderson, Ph.D., Associate Tax- onomist Lewis J. Stannard, Jr., M.S., Assistant Tax- onomist Leonora K. Gloyd, M.S., Laboratory Assistant Philip W. Smith, B.S., Laboratory Assistant Carolyn E. Shroyer, Technical Assistant Section of Aquatic Biology George W. Bennett, Ph.D., Aquatic Biologist and Head WiLLiA.vi C. Starrett, Ph.D., Associate Aquat- ic Biologist D. F. Hansen, Ph.D., Assistant Aquatic Bi- ologist R. VVeldon Larimore, M.S., Research Assist- ant Daniel Avery, Field Assistant Section of Forestry WiLLET N. Wandell, M.F., Forester and Head Lawson B. Culver, B.S., Associate in Forestry Extension Henri D. Crawley, M.F., Junior Forester Technical Library Marguerite Simmons, M.A., M.S., Technical Librarian Ruth Warrick, B.S., Assistant Technical Li- brarian Section of Applied Botany and Plant Pa- thology Leo R. Tehon, Ph.D., Botanist and Head J. Cedric Carter, Ph.D., Plant Pathologist J. L. Forsberg, M.S., Associate Plant Patholo- gist G. H. Boewe, M.S., Assistant Plant Pathologist Robert A. Evers, M.S., Assistant Botanist Joan H. Laube, B.S., B.M., Technical Assistant Section of Game Research and Manage- ment Thomas G. Scott, Ph.D., Game Specialist and Head Ralph E. Yeatter, Ph.D., Game Specialist Frank C. Bellrose, B.S., Associate Game Spe- cialist Harold C. Hanson, M.S., Assistant Game Spe- cialist James S. Jordan, M.F., Assistant Game Tech- nician William Nuess, Laboratory Assistant Cooperative Wildlife Research (Hlinois Department of Conservation and U.S. Fish and Wildlife Service, Cooperating) Paul J. Moore, B.S., Project Leader George C. Arthur, B.S., Project Leader Lysle R. Pietsch, M.F., Project Leader John C. Calhoun, B.S., Assistant Project Leader William J. Harth, M.S., Project Leader Leonard Durham, B.S., Technical Assistant Section of Publications and Public Rela- tions James S. Ayars, B.S., Technical Editor and Head Blanche P. Young, B.A., Assistant Technical Editor Charles L. Scott, B.S., Assistant Technical Photographer Consultant in Herpetology: Hobart M. Smith, Ph.D., Assistant Professor of Zoology, University of Hlinois. This paper is a contributionfrom the .Section of Game Research and Management . (80159—4,500—6-49) <,^g^2 FORE W ORD OF all of the geese inhabiting North America, the Canada goose stands at or near the top of the list in general recognition, and as a game bird. So well known is it that the mention of wild goose brings to the average person a mental pic- ture of the great gray-bodied, black-necked, white-jowled "honker." Francis Kortright, in his The Ducks, Geese and Swans of North America says, "Sagacity, wariness, strength and fidelity are characteristics of the Canada Goose which, collectively, are possessed in the same degree by no other bird." The cold, calculating, investigative, scientific eye may occasionally cast doubt on the completeness with which some of these traits permeate the whole population (as will be noted in this report). One can, if he searches diligently, find a thriftless Scotchman. Wide distribution, great size, and habits conspicuous to the ear and eye have all assisted in making the Canada goose a well-known bird ; but most of the knowl- edge concerning it has been general and superficial. During some time in the year this goose may be seen from one coast to the other and from northern Canada to the Gulf of Mexico. To the average person this wide distribution might mean that the elimination of the species from any of its areas of habitation would be difficult. But every field biologist is familiar with the so- called "flyway concept" that has devel- oped in the past few decades. This con- cept, backed by a large quantity of band- recovery data and general observation, is that the whole population of a migratory species may be divided into subpopulations, each having rather definite nesting and wintering areas and routes of movement. with a minimum of mixing among these suhpopulation groupings. On the basis of this thinking, the study of a migratory species breaks down into a number of geographic units, and the suc- cess or failure of one flyway population may affect but little the populations of other fiyways. This is the problem that faced those in- terested in the geese wintering in Illinois. From nesting grounds on the west side of James Bay in Canada, one segment of the Canade goose population moved south and west, and in recent years wintered to a very large extent at the Horseshoe Lake Game Refuge in Alexander County, Illinois. About half of the population of geese in the Mis- sissippi flyway concentrated in a small area where excessive hunting could conceivably have affected numbers and hunting successes in a very large area both inside and out- side the state. The object of the study re- ported herein was to ascertain the health of the Horseshoe Lake population, and this study required a broad attack both as re- lated to the subject matter investigated and the geography involved. Both of the authors have been far afield in this study. Mr. Smith, as Flyway Bi- ologist for the United States Fish and Wild- life Service, has had an opportunity allowed to but a very few to observe this and other Canada goose populations. Mr. Hanson spent several years at Horseshoe Lake and parts of two summers in the James Bay nesting area The section titled "Population Survival" represents an attempt to analyze a difficult problem with data difficult to obtain in quantity. The data available have been explored by Mr. Hanson, and certain con- clusions reached. These conclusions, it is realized, may vary somewhat from the true picture, but it is felt that their inclusion is worth while as a stimulus to a fuller in- vestigation of this problem even if there were no other values accruing. A study such as the following must of necessity have authors. It is obvious, how- ever, that an investigation of this magni- tude is the result of the authors' efforts plus assistance from many people in numer- ous ways. To all who helped in any way we are deeply grateful. Harlow B. Mills, Chief Illinois Natural History Survey CONTENTS Acknowledgments 67 Materials and Methods 70 Data From Horseshoe Lake 70 Data From Jack Miner Sanctuary 70 Data From (^ther Areas 7.? Data From Questionnaires 7.? The Flyway Concept 74 Eastern Populations 74 North Atlantic Population 77 Hudson-James Bay Populations 77 Hudson-James Bay Breeding Range 79 Limits of Range 79 West Coast Muskey: Types 92 West Coast Production Centers 96 Nest Sites 101 Migration 103 Autumn Migration Routes 10.? Spring Migration Routes 109 Time and Rate of Migrations 110 Winter Concentrations 112 Jack Miner Bird Sanctuary 114 Illinois 116 Michigan 120 Wisconsin 121 Ohio 121 Indiana 121 Arkansas 122 Lower Mississippi River 122 Coastal Marshes 124 Goose Behavior and Hunting Losses 125 Wariness, Innate and Acquired 126 Family Grouping 127 Sociability 128 History of Goose Hunting in Illinois 129 Annual Bag 1.^5 On Breeding Grounds 135 Southern Canada and United States 142 Total Annual Bag 148 Canada vs. United States Kill 149 Differential Hunting Losses 152 Crippling Losses 155 Miscellaneous Mortality Factors 158 Lead Poisoning 158 Starvation 15" Bound Crop 159 Predators 161 Diseases 161 Parasites 162 Productivity 163 Breeding Potential 163 Actual Productivity 166 Data From Horseshoe Lake 166 Theoretical vs. Actual Productivity 171 Flock Sizes 171 Population Survival 172 Definition of Terms 1 72 Mortality 172 Longevity 186 Discussion 1 88 Status 189 Management 191 Present Situation 195 Summary 196 Appendix A, The Southeast Population 199 Breeding Range 199 Migration Routes 199 Wintering Concentrations 199 Future Status 202 Appendix B, Classification of the Canada Geese of the Genus Branta 203 Literature Cited 205 ^ -^ r-' Evening flight of Canada geese at Horseshoe Lake. Canada Geese of the Mississippi Flyway ^^ith Special Reference to an Illinois Flock HAROLD C . HANSON ROBERT H. SMITH* HORSESHOE LAKE, formed from an ancient oxbow of the Alississippi River, lies in Alex- ander County, Illinois, at the southwest tip of the state, fig. 1. An area, totaling 3,489.77 acres, that includes the lake and the island it surrounds, was purchased by the Illinois State Department of Con- servation in 1927 for use as a wildlife refuge. Subsequent purchases in 1941, 1945, and 1946 added about 220 acres to the area, now known as the Horseshoe Lake Game Refuge. That the plan of use for the Horse- s'loe Lake area was eminently successful from the standpoint of attracting wildlife soon became evident. Flocks of Canada geese that previously had wintered along the Mississippi River in the region of southern Illinois left their traditional wintering grounds for the food supph' and the rest lake pro\ ided by the refuge. In recent \ears, for varying periods during the autumn and winter, tiie Horseshoe Lake Game Refuge and the countryside im- mediatel)' around it have contained ap- proximately 50 per cent of the Canada goose population wintering in the entire Mississippi River \ alley. Along with the increase in numbers of Canada geese at Horseshoe Lake there were two developments of pr'mary im- portance: a tremendous increase in shoot- ing pressure on the flock and an altera- tion in the behavior of the geese. Once as wary as any waterfowl population in ' Flyway Biologist, Uniled States Fi^ll and Wildlife Servici the Mississippi Ri\'er valley, the goose flock using Horseshoe Lake gradually lost most of its fear of man and gunfire while near the refuge. The obvious re- sult of the greatly increased shooting pressure and the loss of normal wariness was a tremendous increase in the kill. Large annual kills made at Horseshoe Lake, beginning in 1939, focused the at- tention of wildlife administrators on the need for a long-term management program in that area. In recognition of tiiis need, the Natural History Survey Division of the Illinois State Department of Regis- tration and Education instituted the re- search program on which the present re- port is based. AVhen it became evident that the Horseshoe Lake goose problem was not only of local importance, but national and international in scope, the United States Fish and Wildlife Service initiated a program of investigations to cover the entire range of the Canada goose population wintering in the Mis- sissippi River \alley ; these investigations extended from the James Bay region of Canada to the coastal marshes of Louisi- ana. ACKNOWLEDGMENTS For the generous co-operation on the research project at Horseshoe Lake by the Illinois Department of Conservation, the agency \\hich operates and maintains tiie refuge, special appreciation is due officials or emplo\ees of tiie Department active during the period of field work: [67] 68 Illinois Natural History Survey Bulletin V'ol. 25, Art. .^ HORSESHOE LAKE Fig. 1.—Map showing the boundaries and location of the Horseshoe Lake Game Refuge, 1946. The refuge area totaled about 3,700 acres at the end of that year. The original purchase, in 1927, involved about 3,500 acres. March, 1950 Hanson & Smith: Canada Geese 69 Livingston E. Osborne, Director; Lewis E. Martin, Harold L. Gray, and Joe B. Davidson, members of the administrati\e staff; and Vernal Nave, C. E. Laughery, and Kenneth Long, refuge managers. The inception of the Canada goose study at Horseshoe Lake was in no small meas- ure due to Paul S. Smith, formerly Chief Conservation Officer, Illinois Department of Conservation, and later Game Man- agement Agent of the United States Fish and Wildlife Service. He was one of the first to recognize the serious consequences of the high Canada goose kills in southern Illinois and to awaken others to its dan- gers. The valuable co-operation given the research program by Smith was continued by his successor, Vernon C. Conover. For their official helpfulness and per- sonal interest in the research program, we are indebted to the late Dr. Theodore H. Prison, formerly Chief, to Dr. Leo R. Tehon, formerly Acting Chief, and to Dr. Harlow B. Mills, present Chief, of the Illinois Natural History Survey; to Dr. Clarence Cottam, Fredrick C. Lincoln, Jesse F. Thompson, Leo Couch, Dr. Gustav A. Swanson, and Richard Griffith, all of them during the period of field work with the United States Fish and Wildlife Service; to Dr. Harrison F. Lewis and T. S. Hennessey of the Dominion Wildlife Service, Canada De- partment of Mines and Resources; and to Dr. T. J. Orford of the Indian Affairs Branch, Canada Department of Mines and Resources. Manly Miner, President of the Jack Miner Migratory Bird Foundation, Inc., acting in behalf of the Miner family and the Miner Foundation, was most helpful during the course of this study. For per- mission of the Miner family to compile and analyze the Miner goose-banding rec- ords, we are deeply appreciative. The senior author was privileged to un- dertake field studies in the James Bay area in 1946 and 1947 under the auspices of the Arctic Institute of North America. For helpfulness in connection with this activity we wish to thank Dr. A. L. Wash- burn, Executive Director of the Institute. In 1947, Ducks Unlimited provided the senior author with funds for an aerial re- connaissance of the breeding grounds of the Canada goose in northern Ontario and northeastern Manitoba. Appreciation is due for the support given this program by L. H. Barkhausen, B. W. Cartwright, and Arthur Bartley. Paul Queneau of Westport, Connecticut, rendered valuable assistance during these flights by operating one of the aerial cameras used and keeping a personal flight log of his observations. We are grateful to Father John M. Cooper of the Catholic University of America for permission to quote from his unpublished report on tribal and family hunting grounds in the James Bay area and to reproduce a portion of his accom- panying map, fig. 9. This report, on file in the Indian Affairs Branch, Canada Department of Mines and Resources, was made available to us through the courtesy of T. R. L. Maclnnes. Employees of the Hudson's Bay Com- pany and other residents of the James Bay area have shown us many courtesies and contributed information. W^e are in- debted to William B. Anderson, ^V. J. Cobb, Mathew Cowan, R. M. Duncan, William Faries, C. C. Forman, William Glennie, Patrick Houston, Wesley Hous- ton, A. H. Michell, Father Leopold iMorin, Thomas Rettalack, Norman Ross, Arthur Sullivan, and the late James W'att and Mrs. W^att. The above people, as well as the following, have zealously re- ported bands in past years : J. W^. Ander- son, Bishop Henri Belleau, Father Bilo- deau, George S. Cotter, H. Gibbs, the Rev. Arnold C. Herbert, Brother Gerard Lavoie, the Rev. D. A. MacLachlan, Norman Mathew, L. G. Maver, P. A. C. Nichols, E. H. Riddell, the Rev. H. A. Turner, the Rev. J. H. Turner, J. B. Tyrer, and Harold Udgarden. For information on the wintering grounds and kills of Canada geese in various states we are indebted as follows: for Alichigan, Herbert J. Miller and Dr. Miles D. Pirnie; for Wisconsin, Therman Deerwester and F. R. Zimmer- man; for Minnesota, Frank D. Blair; for Indiana, William B. Barnes; for Iowa. Bruce F. Stiles; for Missouri, M. O. Steen. Data for other states have been gathered largely by the authors, but Frank C. Bellrose of the Illinois Natural His- tory Survey has furnished us with the in- formation on Canada goose concentrations in the Illinois River valley. 70 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 From 1940 through 1943, staff mem- bers of the Illinois Natural History Sur- vey carried out studies of a captive Canada goose flock on the Bright Land Farm near Barrington, Illinois, a program supported by the late A. L. Eustice and Mrs. Eustice, aided by Carleton A. Beckhart. Cecil S. Williams, Dr. Elizabeth Brown Chase, Arthur S. Hawkins, and Dr. Gustav A. Swanson have read parts of the manuscript and given much helpful criticism. Dr. Chase and H. W. Bean have reviewed the statistical data. Arlone Hanson has contributed much valuable assistance in the field and in the office. MATERIALS AND METHODS This Canada goose study is based on data from three primary sources : data collected at Horseshoe Lake, Alexander County, Illinois ; surveys by the authors on the distribution, habitat, and behavior of the population elsewhere in the Mis- sissippi flyway ; and banding records of the Jack Miner Bird Sanctuary, Kings- ville, Ontario. Data from other sources have been used as indicated in the text. Data From Horseshoe Lake Most of the data relating to Canada geese of the Horseshoe Lake Game Refuge, prior to 1940, were obtained by Paul S. Smith when he was federal Game Manage- ment Agent. In 1940 and 1941, Arthur S. Hawkins, then Game Technician of the Illinois Natural History Survey, collaborated with Smith on an investi- gation of conditions at and near Horse- shoe Lake. The first successful trap used at Horseshoe Lake was designed and constructed by John M. Anderson and Jacob H. Lemm of the Natural History Survey and in February, 1941, the first bandings of geese in the area were made by Hawkins, who recorded the sex and age classes of birds banded. In Januarj' and Februarv, 1942, and in the winter of 1942-43, Dr. William H. Elder con- tinued the trapping program begun by Hawkins. From the autumn of 1943 to the spring of 1947, the senior author was responsible for the research program at Horseshoe Lake. In the studies at Horseshoe Lake, par- ticular emphasis was given to trapping and banding (Hanson 1949c), often the only techniques whereby such vital statistics as average longevity and rate of population turnover can be obtained. These study techniques yielded data on sex and age composition of the flock, and, in con- nection with bag inspection, on the dif- ferential vulnerability of the sex and age classes. Sex and age criteria, flock habits and flock organization, crippling losses, and, as time permitted, diseases and para- sites of Canada geese were also studied. The total numbers of Canada geese trapped and banded at Horseshoe Lake by the Illinois Natural History Survey are given in table 1 . Data From Jack Miner Sanctuary On a number of occasions, members of the Illinois Natural History Survey staff Table 1.—Number of Canada geese trapped at Horseshoe Lake, Alexander County, Illinois, by Illinois Natural History Survey personnel, during the fall and winter seasons of 1940-41 through 1946-47. Sf.ason of Trapping .March, 1950 Hanson & Smith: Canada Cjeese 71 Ti9) H U N i)7) BELCHER .ISLANDS X \̂>-^ /Great Whole R. (20) ^^,sV^ (21 [^ l-Eskimo Point 2-Churchill 3-York Foctory 4-Ft. Severn 5-Weenusk 6-Lake River 7- Attawopiskot 8-Akimiski Island 9-Ft. Albany lO-Moose Factory ll-Rupert House LEGEND JAMES m) &*^ .^<>" Afta12-Eastmoin 13-Old Factory 14-Ft. George 15-Greot Wtiale River 16-Port Harrison 17-Povungnituk 18-Gape Smith 19-Wolstenholme ^ 20-Cope Henrietta Mario 21 -Cope Jones 22-Ogoki Ekwon R. ^^ s.0^ BAY ^ Eostmoin B- ,.>!P^ m Fi|5. 2.—Map of the Hiuison-James Bay range of Canada geese that winter in the Mississippi River valley. The main breeding range of this goose population is between the Severn and the Albanv rivers. have visited the Jack Miner Bird Sanc- tuary at Kingsville, Ontario, to study trapping operations. The first traps built at Horseshoe Lake, although set on land, were modeled after the water trap per- fected by the Miners. In May, 1945, the authors visited Kingsville to obtain background material requisite for com- piling and interpreting Miner band-re- covery data. The Miner records con- sisted of the original reports of band re- coveries from hunters in the United States and Canada, and from missionaries and fur traders in the far north, who re- ported recoveries made by the natives. The senior author was responsible for the compilation of these original data, which are filed in Ottawa at the Dominion Wildlife Service, Canada Department of Mines and Resources. JJetween 1915 and the spring of 1944, approximate!)' 31,000 Canada geese were banded at the Miner Sanctuary. From these bandings approximately 3,900 rec- 72 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Table 2.—Number of Canada geese banded at the Jack Miner Bird Sanctuary, Kingsville, Ontario, during the fall trapping seasons, 1927-1944. Year March, 1950 Hanson & Smith: Canada Geese ,^to"^ ^^°>-^ 73 ^- %^ i^' ^^ HUDSON BAY h— •:X.yiii ,^^' ^^',6^^ Winj.sk R. \IV _Ekwan , ->-i'-',^- ^ JAMES BAY •FT. ALBANY IIV OGOKI- *»" ^-0" ic:'^ ^ Fig. 4.—Map showing flight routes of an aerial reconnaissance of the breeiling grounds of Canada geese south of Hudson Bay and west of James Bay in 1947. Roman numerals designate the various flights discussed in the text. Recovery records for the first 8 years are incomplete, as many of the letters re- porting bands were given to newspapers and never returned. In some cases only news clippings with incomplete data served to preserve early records. Data From Other Areas Field studies on Canada goose concen- trations away from Horseshoe Lake were begun by the junior author in 1942. Be- ginning in 1943, he inventoried by plane many of the wintering concentrations from Horseshoe Lake to Louisiana. He de- voted the summer of 1943 to a survey of the south and east coast areas of James Bay, from Moose Factory, Ontario, to Fort George, Quebec, fig. 2. The follow- ing summer he made a reconnaissance of the west coast from Moose Factory to Cape Henrietta Maria. The senior author made a brief pre- liminary trip to James Bay in the summer of 1946, visiting Moose Factory, Rupert House, and Fort Albain, and ascending Little Partridge Creek via canoe. In 1947, he spent from mid-May to Septem- ber investigating the breeding grounds in- land from the west coast of James Bay ; he used both canoe, fig. 3, and plane for these surveys. The aerial reconnaissance in 1947 included stops at Weenusk, Fort Severn, and York Factory. Approxi- matelv 375 aerial photographs were taken on this aerial survey, the itinerary of which is shown in fig. 4. Data From Questionnaires Approximate'y 40 questionnaires re- garding goose-breeding grounds and kills were distributed to fur trade posts in the Canadian Eastern Arctic in 1947, througii the courtes\ of the administra- tion of the Northwest Territories, Canada Department of Mines and Resources. Replies to the questionnaires have been summarized and the data included in this report. 74 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 THE FLYWAY CONCEPT Over a decade ago, Lincoln (1935) presented the concept that the routes taken by North American birds in migra- tion fall into major flyways or lanes of travel. Recoveries of banded birds have demonstrated the validity of the flyvvay concept with respect to waterfowl as well as many other kinds of birds. Lincoln named the Atlantic, the Mississippi, the Central, and the Pacific flyways as the principal ones of North America. The limits of the waterfowl flyways vary somewhat with each species and may change to some degree from year to year, depending on weather, surface water, and food conditions. In most species the populations of one flyway merge al- most imperceptibly with those of ad- joining flyways. Consequently, the fly- way taken in any one year by an individ- ual bird breeding in an area where two flyways meet may be due in part to chance. The adherence of ducks and geese to their ancestral flyways has been demon- strated experimentally by removing in- dividuals from one flyway to another. With relatively few exceptions, the trans- ported individuals have been recorded later in their original flyways. One of the early experiments of this kind with ducks was begun in 1918 by Mcllhenny (1940), who, in co-operation with Dr. Arthur A. Allen of Cornell University and the United States Bureau of Biolog- ical Survey, shipped ducks and coots trapped during the winter in Louisiana, which is in the Mississippi flyway, to points in the Atlantic and Pacific fly- ways. Most of the released individuals that were later recovered or retrapped were taken in the Mississippi flyway. Perhaps the earliest test of this kind with Canada geese was made by Jack Miner; complete data on the test were found in the files of the Dominion Wild- life Service. In the spring of 1934, 25 geese trapped at the Miner Sanctuary, from flocks that had wintered on the Atlantic Coast and were in migration to their breeding grounds along the east coast of James and Hudson bays, were released among a concentration of blue and snow geese at Grant Lake, Mani- toba, a locality far west of their own migration routes. Three of these geese were later reported shot, two of them in their own flyway: one in the vicinity of Poplar Branch, North Carolina, in the fall of 1934; the other near Lake St. John, Quebec, in the fall of 1940. The third was recovered in northern Mani- toba in the spring of 1934, too soon after release for the record to be significant. The chief deviations from flyway con- sciousness are among young birds that have not yet nested (Lincoln 1934). Williams & Kalmbach (1943) showed that the migratory behavior of young Canada geese when raised in or trans- ported to a new area is similar to the be- havior of geese native to that area. As pointed out by Lincoln (1935), the adherence of waterfowl to their ancestral flyways has particular administrative significance in connection with conserving the continental waterfowl resources. "It indicates," Lincoln writes, "that if the birds should be exterminated in any one of the four major flyways now defi- nitely recognized, it would at best be a long time before that region could be re- populated, even though birds of the species affected should continue over other fly- ways to return to their great breeding grounds of the North." This hypothesis is of special significance as applied to the management of Canada geese. Members of a species with a fairly low breeding potential, they would prob- ably require several years to regain their numbers in any one flyway after having been once seriously depleted. Thus, it is to the hunter's best interests that the yearly kills in each flyway be kept within reasonable bounds. EASTERN POPULATIONS A brief review of the distribution and taxonomy of Canada goose populations in eastern North America is relevant to an understanding of the data later pres- ented concerning the Mississippi flyway population. The Canada geese using the Atlantic and Mississippi flyways, as defined by Lincoln (1935), have been recognized as belonging to two distinct major popula- tions, based on taxonomy (Todd 1938) March, 1950 Hanson- &: Smith: Canada Geese 75 Fig. 5.—Extreme examples of plumage variation in Canada geese of the flock wintering at Horseshoe Lake in southern Illinois. The majority of the geese at Horseshoe Lake approach the dark-colored goose, left above, considered to be Branta canadensis interior; but a few resemble the individual at the right. The latter is more like Branta canadensis canadensis of the North .•\tlantic coast. The goose on the left is a yearling female; that on the right, a yearling male. and location of the breeding grounds : the North Atlantic population and the Hud- son-James hay population. The North Atlantic population constitutes a distinct management unit. The study reported here indicates that the Hudson-James bay population is not homogeneous but con- sists of four subpopulations, each of which constitutes a separate management unit having a fairly distinct range of its own. These subpopulations are here designated by terms suggestive of their wintering grounds or migration routes : the South Atlantic, the Southeast, the Mississippi \'alley, and the Eastern Prai- rie.* The ranges of these subpopulations are shown in fig. 6. Todd (1938) noted what he considered significant plumage dififercnces among • Name and recognition of the Eastern Prairie popula- tion as a separate population from Cecil S. Williams of the United Stales Fish and Wildlife Service. 1946. Canada geese collected in the eastern lialf of the United States and proposed a new subspecies, Branta canadensis in- terior, for the darker colored birds that breed and migrate in an area west of the range of the nominate subspecies, Branta canadensis canadensis. Fig. 5 shows two Canada geese trapped at Horseshoe Lake with plumages that illustrate some of the differences between these two races. " Typical canadensis, as represented by breeding examples from Newfoundland and by winter birds from the South Atlantic coast, is a comparatively light- colored bird," according to Todd (1938). "In breeding dress the anterior under parts are buffy white, and this pale color runs up on the sides of the lower neck (behind the black) to form a conspic- uous light-colored area on the upper back. In the new race this feature is wanting. The feather-edgings of the new 76 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 FLYWAY POPULATIONS ii?g* EASTERN PRAIRIE MISSISSIPPI VALLEY SOUTHEAST SOUTH ATLANTIC Fig. 6.—Map showing roughly the main ranges of the four populations of Canada geese nesting in the Hudson-James bay region. The range of the JVIississippi Valley geese overlaps the range of the Southeast population chiefly in fall; the range of the South Atlantic popula- tion overlaps the range of the Southeast population chiefly in spring. The western limits of the range of the Eastern Prairie population extend farther west than indicated here. The eastern limits of the range of the South Atlantic population probably extend farther east in some areas than indicated. March, 1950 Hanson & Smith: Canada Geese 77 race are generally darker, while the un- der-plumage is conspicuously so." Official recognition was given to the race interior by its inclusion in the Twentieth Supplement to the American Ornithologists' Union Check-List of North American Birds (Wetmore 1945). Appendix B contains a brief summary of the latest classification of the Canada geese of the genus Branta, with notes regarding recognition of various kinds by the Indians. North Atlantic Population The Canada geese of the North Atlantic, which breed in Newfoundland, eastern Quebec, and Labrador north to the northern limit of trees (Austin 1932), are those recognized by Todd (1938) as Branta canadensis canadensis. In the autumn, they migrate down the Atlantic Coast and winter principally from Port Joli and Port I'Hebert, Nova Scotia (Tufts 1932. Lloyd 1923), to Martha's N'ineyard, Massachusetts, and south prob- ably as far as New Jersey. Skins ex- amined by us at the Chicago Natural Histor)' Museum indicate that some of these geese winter as far south as the coast of North Carolina, where they mingle with South Atlantic geese. Low (1935), in a report on 64 Canada geese banded at Cape Cod, Massachusetts, presented convincing evi- dence that the flight of geese along the North Atlantic Coast is a distinct entity. Twenty-five of the 26 geese later re- covered or recaptured were taken between Newfoundland and New Jersey. One was recovered in Florida. Hudson-James Bay Populations The Canada geese that breed inland from both coasts of Hudson and James bays, fig. 2, as far north on the west coast as Churchill, Manitoba, and prob- ably as far north on the east coast as Bafifin Island, which lies just north of Cape ^Volstenholme, conform to the description given by Todd (1938) for Branta canadensis interior. While the distribution of geese breeding around the two bays is more or less continuous, available data indicate that this popula- tion is a heterogeneous one and is com- posed of the four segments or subpopula- tions pre\ iousl> named: the South Atlan- tic, the Southeast, the Mississippi Valley, and the Eastern Prairie. Each has its own breeding range, migration routes, and wintering areas, figs. 6 and 7. The exist- ence of two of the population divisions that nest in the Hudson Bay area, one wintering along the central Atlantic Coast and the other in the Mississippi River valley, was first pointed out by Manly Miner (1931). This discovery, based on band recoveries, was due in part to the fortuitous location of the Miner Sanctuary, fig. 12, which lies about midway between the migration routes of these populations and thus per- mits banding of both populations. South Atlantic Population.—This population is distributed in winter along the Atlantic Coast from southern New Jersey to Chesapeake Bay, Back Bay (Virginia), Pamlico Sound, and Curri- tuck Sound, and Hyde and Dare counties, North Carolina. Recoveries from geese migrating through the Miner Sanctuary in the spring, and banded there in that season, reveal that Lake Mattamuskeet in Hyde County, North Carolina, has in recent years become the most important wintering area of this population. A portion of the birds in this popula- tion stop at the Miner Sanctuary while en route to their breeding grounds, which are on the Belcher and prohabh the Twin Islands and in suitable localities along the east coast of James and Hudson bays, and inland probably to the height of land, as suggested by Todd (1938). Band recoveries indicate that the breeding range may include a portion of southern Baffin Island, fig. 7. Large numbers of reco\eries reported from a post or small area may actually have been taken along extensive areas of the coast. For instance, recoveries plotted as from the Belclicr Islands in fig. 7 also include the recoveries from the east coast of Hudson Bay from Cape Jones to Nastapoka Falls ; recoveries represented as from the Port Harrison area actually include the recoveries made along the east coast of Hudson Bay from the Kikkcrteluk River area to the Povung- nituk area. Southeast Population.—The exist- ence and range of the Southeast popula- tion was revealed when band recoveries Illinois Natural History Survey Bulletin Vol. 25, Art. 3 EASTERN PRAIRIE MISSISSIPPI VALLEY SOUTHEAST POPULATION POPULATION POPULATION SOUTH ATLANTIC POPULATION 228 RECOVERIES OF GEESE BANDED AT HORSESHOE LAKEJLLINOIS N ez'-iiiZl March, 1950 Hanson & Smith: Canada Geese 79 trom the Miner autumn-banded geese were plotted as to exact locality. These recoveries show that the autumn flight of geese through the Kingsville, Ontario, area is not homogeneous, but is composed of two populations of geese : the South- east population and the Mississippi \'alle\ population, tigs. 6 and 7. The Southeast population breeds inland from the south coast of James Bay and winters in the in- land regions of the southeastern states. A detailed discussion of the range of the Southeast population is presented in Ap- pendix A. Mississippi Valley Population.— The range of the Canada goose popula- tion that winters in the valley of the Mis- sissippi River extends in autumn and winter from western Michigan west through the eastern portions of those states lying immediately west of the Mis- sissippi River and south in the valley of this river to the coast of the Gulf of Mexico. The main winter range south of Cairo, Illinois, does not extend great- ly beyond the immediate valley of the Mississippi River except in Arkansas and Louisiana. The Mississippi Valley popu- lation, which is gi\en primary considera- tion in this paper, breeds inland from the west coast of James Bay and the south coast of Hudson Bay, figs. 6 and 7. Eastern Prairie Population.—The eastern range limits of the Eastern Prairie population seemingly merge with the western range limits of the Mississippi \'alley population on the breeding grounds in the muskeg between Fort Severn and Fort York and on the wintering grounds in western Louisiana, figs. 6 and 7. The eastern range limits of the Eastern Prairie geese in migration are apparently in cen- tral parts of ^Minnesota, Iowa, Missouri, Arkansas, and Louisiana. We do not have the data at hand to discuss the western limits of the range of this popu- lation, nor are the\- of concern in this paper. HUDSON-JAMES BAY BREEDING RANGE The Canada goose has long been a staple food item for the natives of North Ameri- ca. To the white man in the United States and Canada, it has been a highly prized hunting trophy as well as an esteemed table bird. Formerly the species nested over much of the upper Mis- sissippi River valley (McClanahan 1940), but, subjected to intensive hunting pres- sure, it was soon extirpated as a breeding bird from most of this country. Prob- ably the only reason that there are still Canada geese to winter in the Mis- sissippi River valley is that much of the country adjacent to Hudson and James ba\s in northern Ontario, where most of tliis migratory population breeds, is rel- atively inaccessible to man in summer. Limits of Range The general limits of the range of the Canada goose in the Hudson-James bay area have not been adequately summarized in previous publications. The existence of only two of the four populations that nest adjacent to these bays has been rec- ognized previously, and the limits of their ranges have not been well defined. For these reasons, in addition to presenting new data on the Canada goose breeding range in the region of Hudson and James bays, we review pertinent references in the literature. Until the race Braiita auiadtnixis in- terior was recognized by the American Ornithologists' Union (Wetmore 1945), most of the writers who mentioned the Canada goose either made no distinction between the two races of Branta cana- densis, or they referred to birds of both races as belonging to the race canadensis. References in the literature prior to 1945 to either of these races should be inter- preted in the light of the recent decision by the A.O.U. The sequence of the following citations is in general according to the geographic position of the localities concerned : from north to south on the west side of the bays and from south to north on the east side. The northern limit of the breeding range of Branta canadensis interior west of Hudson Bay coincides roughly with the northern limit of trees as delineated by the distribution of black spruce and white spruce, fig. 8. Taverner & Sutton (1934) found that at Churchill, Manitoba, which is "precisely at the limit of tree growth, where the spruce forest dies out on the 80 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Fig. 8.—Map showing subsurface geological structures south and west of Hudson and James bays. Approximate limit of trees from official Canadian map. The principal nesting range of the Mississippi Valley Canada goose population lies within the shaded area. arctic tundra and both types of biological association are in contact," the goose they referred to as Branta canadensis cana- densis "is a common transient, which breeds sparingly in the vicinity." Preble (1902) recorded that when he was in the region west of James Bay and Hudson Bay considerable numbers of Branta canadensis were reported as nest- ing on an island in Lake Winnipeg. He saw or had reliable reports of young geese along the Fox, the Churchill, and other rivers of the region. Bell (1880) stated that Jnser cana- densis "breeds in considerable numbers along the Churchill River." Grinnell & Palmer (1941) reported that "birds [Branta canadensis^ were seen and heard at intervals from June 6 on" in the vicinity of Churchill. Allen (1945) recorded nests and nest- ing pairs of Branta canadensis near Churchill. Traverner (1931) wrote: ". . . . it [Branta canadensis candensis] is the common breeding goose of James and Hudson bays for most of the east coast and the west side at least as far as March, 1950 Hanson & Smith : Canada Geese 81 Churchill, probably stopping somewhere south of cape Eskimo where it appears to be replaced by leucopareia." From the barren grounds north of Churchill, the lesser Canada goose, Branta leuco- pareia leucopareia is the common repre- sentative of the genus Braiila.* Herring (1937) examined parts of 10 individuals, mainly from Baker Lake, a locality 385 miles nearly due north of Churchill, which he assigned to the race leucopareia. A female goose taken May 20, 1937, at Eskimo Point, on the coast of Hudson Bav north of Churchill, appeared to Shortt & Peters (1942) "to be of the form leucopareia." Specimens taken along the Thelon River, which is in the districts of Mac- kenzie and Keewatin, Northwest Terri- tories, were "referred by P. A. Taverner to B. c. leucopareia" (Clarke 1940). There are numerous references in the literature regarding the occurrence of the Canada goose south of Hudson Bay and west of James Bay. Richardson (1851) quoted a report of George Barnston, an officer of the Hudson's Bay Company at "Martin's Falls," a post on the Albany River 200 miles inland from James Bay, in which mention is made of "geese and ducks hatching" in the vicinity. Bell (1887), describing his exploration of the Attawapiskat River, wrote: "The Canada goose breeds in considerable num- bers in the open swamps behind the wooded borders of the lower section of the river, and the young birds, ready to fly, were congregating in flocks, all along the lower stretch, in the end of August and the beginning of September." Baillie & Harrington (1937) wrote: "The Canada Goose breeds fairly com- monly along the coasts of James and Hudson bays, between Moose river and Churchill." South of James Bay the principal breed- ing range of the Canada goose may not extend more than 60 miles inland from the coast. In 1926, a mining party led by B. C. Lamble explored the country between Timmins, Ontario, and James Bay. During the trip they "saw many broods of Canada geese, but none farther • See Appendix B for discussion of recent revision by Hellmayr & Conover (1948). south than Kesagami Lake, Latitude 50° 30'" (letter to Jack Miner from B. C. Lamble, August 5, 1926). In regard to the status in other parts of Ontario of the bird they regarded as Branta canadensis canadensis, Baillie & Harrington (1937) stated: "Recent maps indicate, perhaps correctly, that this bird may breed in the whole of northern On- tario, north of Lake Superior and the southern end of James Bay. "The several instances of this bird nesting in southern and central Ontario almost undoubtedly concern injured or semi-domesticated individuals." Bell (1883), who was undoubtedly in- timately familiar with most of the prov- ince of Ontario, stated that "between the great lakes and James' Bay, only chance pairs lag behind in their north- ward flight to hatch their broods." Inland from many parts of the east coast of Hudson and James bays, and on the islands along the coast and to the north, suitable habitat for nesting Canada geese is less extensive than inland from the west coast. Consequently, nesting on the east side of the bays is relatively con- centrated although, in the interior of northern Quebec (Ungava), more widely scattered nesting is found. The late James Watt, former manager of the Hudson's Bay Company post at Rupert House, wrote the junior author (letter of December 25, 1943) that "While travelling in the interior [south and east of James Bay] surveying beaver lands and counting lodges I have seen as many as 15 to 20 nesting [Canada] geese in a day's travel—all with broods of young geese, and .... taking into consideration the immense territory and number of lakes and inland waterways, the number of geese that nest inland must be large." A. P. Low (1896) wrote: The Canada goose "breeds in marshes throughout the northern interior [of Quebec], and is seen along the rivers with young broods about July 1st; several large broods seen on Burnt Lakes, Romaine River ; not common at Lake Mistassini, but abundant on East Main River— • especially on lower part, where the river is cut out of clays, with good bottom- lands; breeds in large numbers on the islands of James Bay." 82 Illinois Natural History Survey Bulletin Vol. 25, Art. .3 In reply to the questionnaire sent out in 1947, Roy Jefferies, Post Manager at Eastmain for the Hudson's Bay Com- pany, in collaboration with an Indian, stated that Canada geese nested in a swamp about 10 miles south of East- main. It is common knowledge in the James Bay area that considerable concentrations of nesting geese are found on the Twin Islands in James Bay, particularly the South Twin, and on the Belcher Islands in Hudson Bay, fig. 2. Nesting pairs are also found on Charlton Island and a number of smaller islands along the east coast of James Bay. In the summer of 1947, Donald F. Coates and Donald B. Coombs (personal communication), observers for the Geodet- ic Service of Canada, visited the follow- ing islands in James Bay : North Bear, Bear, South Bear, Bare, Grey Goose, Walter, North Twin, Weston, and Charl- ton. They found Canada geese on only three of these islands. On Grey Goose, their guide shot two geese but they found little evidence of breeding pairs ; on Wes- ton, they saw about 20 pairs, in one instance 6 adults and 21 goslings to- gether on one pond ; and on Salt Lake, at the northern tip of Charlton, they ob- served 1 pair and 6 goslings. Bell (1883) found that Canada geese "breed on the islands along the east coast of Hudson's Bay it is said that very few Canada geese breed north- ward of Hudson's Strait." Manning (1946) mentioned "a con- siderable number of geese in the Mistake Bay area at the end of July." (Mistake Bay is between Povungnituk and Port Harrison, fig. 2.) He "saw 10 or 15 of them, and all belonged to the large form." He identified them as Branta canadensis interior. In a recent letter (to the senior author, April 11, 1947), T. H. Manning states that he believes the chief breeding ground of Canada geese in this area is between Cape Dufferin (near Port Harrison) and the Cape Smith Range. "I do not . . . think that they often nest on the coastal islands. They may nest on the King George and Sleeper Islands, but the Ottawa Islands are high, rocky and bar- ren, and unsuitable. I have no direct evi- dence, but I should think they nest throughout the interior between Hudson and Ungava bays." The Reverend H. S. Shepherd, mis- sionary at Port Harrison, Quebec, for 2 years, stated in the questionnaire sent out in 1947 that scattered nesting of Canada geese is found over a large area of the interior inland from Port Harrison, but that the total number is not great. Low (1902) found that, in the country about 12 miles south of the Digges Islands,* "The many small ponds and swamps that occur between the boulder ridges are favorite breeding places for grey geese." Farther south, about 30 miles north of the Povungnituk River, Low also found large numbers of Canada geese about 10 miles inland from the mouth of the Sorehead River. In the Povungnituk area, W. A. Tol- boom, a post manager, reported by questionnaire in 1947 that nesting is well scattered over a wide area and that generally speaking all nests are found on islands, on lakes or shores of lakes, seldom on rivers, and very seldom on coastal islands. Manning's surmise regarding goose nesting over the Ungava Peninsula is substantiated by Rousseau's ( 1948) finding that the Canada goose is one of the few prevalent forms of wildlife between Povungnituk and Payne Bay post on Ungava Bay. A few individuals of the Canada goose nest on the arctic islands north of the Canadian mainland. Sutton (1932) re- ported that Eskimos have occasionally found nests of Branta canadensis cana- densis on Southampton Island. Shortt & Peters (1942) reported an immature "specimen referable to B. canadensis canadensis" taken August 17, 1938, at Lake Harbour, southern Baffin Island. Soper (1946) reported that Branta canadensis canadensis breeds on Baffin Island along the southern coast of Foxe Peninsula, and from at least Amadjuak Bay to Gabriel Strait along the coast of Hudson Strait. Mississippi Valley Population.— The limits of the breeding range of each * Small islands lying off the extreme tip of Ungava, northern Quebec, March. 1950 Hanson & Smith: Canada Gekse 83 of the four populations of Branta cana- densis interior around Hudson and James bays have been deduced from band re- coveries coupled with a knowledge of the suitable nesting country, Hgs. 6 and 7. In ascertaining the true distribution of each of these populations, we were for- tunate that the Miner banding records, as well as the Horseshoe Lake records, could be analyzed. An interpretation of either the Horseshoe Lake records or the Miner records alone would undoubtedly have led to erroneous conclusions, wjiere- as the two sets of data considered to- gether supplemented each other. Band-recovery data from the Hudson- James bay area are in large measure de- pendent upon the native Indians. When interviewed through interpreters, the Indians are usually able to furnish the exact date and place of each band re- covery. However, if only the name of tlie post is known at which a band is se- cured, the location of the recovery can be approximated, as usually the native groups from the various fur trade posts, including even individual families, use the same liunting grounds year after year. The approximate boundaries of the hunt- ing grounds of the various bands of Indians on the west and south coasts of James Bay is shown in fig. 9, ^\•hich is copied from a portion of a map prepared by the Reverend John M. Cooper to ac- LEGEND LIMITS OF HUNTING TERRITORIES O TRADING POST AND INDIAN BAND Fig. 9.—Map showing limits uf the trapping and hunting grounds of the various bands of Cree Indians west and south of Hudson and James bays. (After Cooper 1933.) 84 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 March, 1950 Hanson & Smith: Canada Geese 85 compan\- his unpublished report to the Indian Affairs Branch in Ottawa. Father Cooper's map is based on his field studies of 1927, 1932, and 1933 in the vicinity of James Bay. According to Father Cooper, the boundary lines of hunting territories generally are represented by natural land- marks, such as heights of land, chains of lakes, or watersheds. Hunting rights to each foot of land are owned by some Indian and are acquired only through in- heritance or donation. Territory is generally inherited in the male line and most family territories have been in the same family line for generations. As to the accuracy of these boundaries, Father Cooper writes : "These limits are and can be only appro.ximate as we have not ade- quate and detailed maps based on surveys of the whole area The Moose Indian grounds and to a certain extent the Rupert House grounds are plotted as of a generation ago. Some changes through inheritance and through the dying out of certain families, particularly around Lake Kesagami,' have occurred, but in the main the present Indian families still hunt each where the father and grandfathers hunted." Band recoveries from the Canadian breeding grounds of geese banded at Horseshoe Lake are summarized in table 3. These recoveries, important in re- vealing the location and extent of the breeding range of most of the Mississippi Valley geese, do not, however, take into account geese that nest in the United States, where several efforts to establish breeding flocks on federal, state, and private refuges are making increasingly important contributions to the Mississippi Valley population. Most of the Horseshoe Lake bands recovered in Canada were taken in the muskeg country lying inland from the coasts of James and Hudson bays between the Kinoje* and Severn River watersheds, fig. 30. Band recoveries indicate that during the breeding season this enormous section of muskeg country, roughly tri- angular in outline, contains the bulk of the geese that winter in southern Ontario, Michigan, Wisconsin, Indiana, Illinois, • A small river that flows into James Bay 8 miles south of the Albany River. eastern portions of Minnesota, Iowa, Missouri, Arkansas, Louisiana (Delta), western Kentucky, western Tennessee, and western Mississippi. The barren grounds of Cape Henrietta Maria and the coastal marshes probably do not con- tain breeding birds, but nesting occurs on Akimiski Island, figs. 2 and 30, which lies in James Bay a few miles east of the mouth of the Attawapiskat River. Of the tremendous area of muskeg outlined above, only a relatively small por- tion is either suitable for, or attractive to, nesting geese. Field observations, as well as band recoveries, indicate that the main breeding range of Branta canadensis in- terior south of Hudson Bay and west of James Bay is within an enormous area of muskeg, the limits of which coincide roughly with the area underlaid with sedimentary rocks of the Paleozoic era, fig. 8. These rocks, of the Ordovician, Silurian, and Devonian periods, are covered by a mantle of glacial drift over which the flat muskeg is superimposed. According to Bell (1887), "The drift (principally boulder-clay) which over- spreads the palaeozoic basin westward of James' Bay appears to be a continuous sheet varying probably between thirty and ninety feet as far as can be judged by the sections along the rivers." Ells (1912) believes that fairly uni- form timber and land conditions prevail concentrically from James Bay except for minor variations, depending on primary and secondary drainage. Thus, if we have a 5 ft. muskeg at a distance of thirty miles south of James Bay, I would look for a similar condition East and West along a belt roughly parallel with the shores of the Bay This as- sumption I have based on the fundamental principle that the country adjacent to James Bay on the South and West side is grad- ually being elevated As we leave the shores of James Bay, the depth of the muskeg should gradually increase .... Eight miles to the west of Moose Factory the depth of moss and muck is 2 ft. to 3 ft.; 10 miles further south the depth is 2 ft. to 4 ft. ; and 40 miles, 4i/i to 5 ft. ; at 60-80 miles, 5yz to 6 ft.; and at 90 miles the depth is 6 to 8 feet. Evidence that the main breeding grounds of the Mississippi Valley Canada Illinois Natural History Survey Bulletin Vol. 25, Art. .1 SCALE IN MILES 50 100 200 NUMBERS REFER TO THE SURVEYING PARTIES Fig.* 10.—Map showing districts explored by surveying parties in northern Ontario in 1900 (from Anonymous 1901). None of the surveying parties reported nesting Canada geese in the districts surveyed. geese probably do not lie much beyond the coastal strip of country that is under- laid with sedimentary rock, and are not in the adjoining rocky and rugged Cana- dian shield, is found in the records of 10 surveying parties. In the summer and autumn of 1900, surveys and explora- tions were made of the natural resources and characteristics of part of northern Ontario by the Ontario Department of Crown Lands (Anonymous 1901). The subject of the survey was a "comparatively unknown part of the District of Nipissing, bounded on the north by the Great Mus- keg, adjoining the southern shore of James' Bay." The country, beginning about 80 miles inland from James Bay, was surveyed by districts, an exploring party being assigned to each of 10 dis- tricts. The districts that have relation to this study are shown in fig. 10. Each exploring party kept notes on the game conditions in its respective district, and, although a number of kinds of ducks were reported nesting in several of the districts, no nesting Canada geese were noted. Observations made by Hess (1943) during a plane flight shed additional light on the occurrence of Canada geese south- west from James Bay. His description of the muskeg in that sector could apply to a large portion of the muskeg over the Paleozoic Basin. By the time we were in McCausland Township, the country had changed from the poplar and jackpine regeneration on the slopes around the Mattagami River to a vast flat area of muskeg, exactly similar to the country around James Bay. (At this point, we were about 100 miles from the Bay.) Throughout this area, except for a belt of fair-sized spruce along the rivers and larger streams and the bigger lakes, there was no tree growth except dwarf widely-spaced tamaracks and the odd bunch of black spruce trees. The remainder of the area was a greyish yellowish green blanket of moss interspersed in large patches by ripple-like depressions filled with water. March. 1950 Hanson ^^ Smith: Canada Cii-Esi-: 87 giving a striking similarity to waves of moss and water. Hess observed only four geese in the area described above. After flying a con- siderable, but unstated, distance farther, he sighted a chain of five lakes, on one of which were two flocks of geese ; "\oung appeared to be present." Hess reported these lakes as being shallow. The shores to about 100 feet from the water are ringed by black spruce trees about .^0 feet high which shade off a short distance from the lake into the muskeg. The im- mediate shore was covered bv alder and willow and in the far distance the larger spruce trees along the Missinaibi River stood out sharply above the scrub larch and muskeg. Other Populations.—From the 5.747 Canada geese banded at Horseshoe Lake, only 4 bands have been recovered from the country adjacent to Hudson Bay northwest of Fort Severn, fig. 30. Two of these bands were from geese killed in early spring south of ^'ork Factory, ap- parently in the vicinity of their breeding grounds. Of the more than 16,000 Canada geese banded at the Miner Sanc- tuary in the autumn, none has been re- Fi<. 11.—Location of band recoveries from Canada geese banded at the Jack Miner Bird Sanctuary in the autumn, 1915—1944, and reported recovered south of James Bay in Canada. Recoveries reported from fur-trade posts on the coasts of Hudson and James bays are indicated in fig. 7. Banding records indicate that two Canada goose populations, the Mississippi Valley and the Southeast, stop at the Miner Sanctuary in the autumn. 88 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Fig. 12.—Location of band recoveries from Canada geese banded at the Jack Miner Bird Sanctuary, Kingsville, Ontario, in the spring of years prior to 1934 and reported recovered in the United States in the autumn and winter of 193 5-36 and (one from a goose found dead) in the spring of 1936. ported shot north of Lake River, figs. 2 and 7. Hence, it seems probable that most of the muskeg country between Fort Severn and Churchill is occupied by a population of geese that by-pass both the Miner and Horseshoe Lake refuges on their migration to wintering quarters. Miner bands recovered in Canada south of James Bay are indicated in fig. 11. Since few geese banded either at Horse- shoe Lake or the Miner Sanctuary have been reported from western Louisiana and eastern Texas, or from any point at an appreciable distance west of the Mis- sissippi River, figs. 12—21, it appears that the western Louisiana flocks, and perhaps a few concentrations in central Missouri, are derived from the breeding grounds between Fort Severn and Churchill. (Fig. 12 shows recoveries of geese banded in the spring; figs. 13-21 show recoveries of geese banded in the fall and winter.) The geese that breed in this part of Canada should probably be included with the Eastern Prairie population, as recently proposed by Cecil S. Williams of the United States Fish and Wildlife Service. It is apparent from the distribution of the band recoveries shown in figs. 14—21 that two distinct populations of geese are banded at the Miner Sanctuary in the autumn : one population that migrates southwest to winter in the Mississippi River valley; the other, designated as the Southeast population, fig. 6, that crosses the Appalachian Mountains and winters in the inland areas of the South Atlantic states. From the data at hand we can only speculate on the approximate line of demarcation between these two popula- tions on the breeding grounds. Although band recoveries indicate that the breeding grounds of the Mississippi flyway popula- tion extend as far south as the Kinoje River, the mouth of which lies 8 miles south of the mouth of the Albany River, between the Kinoje River and the Moose River country there may be a zone of overlap in which is found a mixed popula- Fig. 13.—Location of band recoveries from Canada geese banded at the Horseshoe Lake Game Refuge and reported recovered in the United States and southern Ontario, 1940- 1945. (Missouri recoveries near Horseshoe Lake.) March, 1950 Hanson t^- Smith: Canada Geese 89 YEAH OF RECOVERY * 1925 * I^ HOLLOW SYMBOLS SHOW IMS LOCALITY IS DOueiFUL T 1929 Fig. 14.—Location of band recoveries from Canada geese banded at the Jack Miner Bird Sanctuary during the autumn, 1925-1929, and reported recovered in the United States during the season of banding. tion of geese, some of which winter in the Mississippi River valley and others that winter in the inland portions of the South Atlantic states. The principal breeding range of the Southeast population lies inland from the south coast of James Bay. The data available at present suggest that it in- cludes areas drained by the ]\Ioose River, as well as suitable muskeg lying between the Moose and Nottaway rivers, and perhaps areas lying inland from the east coast for an indeterminate distance north, fig. 6. Banding records from the Miner Sanctuary show that many of the autumn- banded geese are taken in the spring in the country around the south end of James Bay, fig. 7, and many are taken in the autumn in the inland portions of Virginia, North Carolina, South Caro- lina, Georgia, Alabama, and the Gulf Coast of Florida, figs. 1-1—21 (also see Appendix A). Recoveries from geese banded at the Miner Sanctuary in the spring clearly indicate that the breeding grounds of the flocks that winter along the Atlantic Coast from Maryland to North Carolina, fig. 12, include certain islands in James and Hudson bays (see pages 81-82) and areas inland from the east coast of these bays from about Rupert House to southern Baffin Island, fig. 7. The large number of band recoveries from the Port Harrison region on the east coast of Hudson Bay, despite low nesting densities reported for that area, may be due in part to the influ.x of geese in late summer into this lake country, which lies north of the tree line. Accord- ing to the Re\erend H. S. Shepiierd, large numbers of Canada geese fly in from the north to the barren-ground lakes for the purpose of moulting. Band re- coveries suggest that there may also be an influx of geese that have flown in from considerably south of Port Harrison. No confirmation of this influx was received in the questionnaire distributed in the re- gion; however, A. Lunan of the Hudson's YEAH OF RECOVERY A 1930 " t^l HOLLOW SYMBOLS SHOW '932 LOCALITY IS OOUeTFUL 1933 1934 Fig. 15.—Location of band recoveries from Canada geese banded at the Jacli Miner Bird Sanctuary during the autumn, 1930-1934, and reported recovered in the United States during the season of banding. 90 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 YEAR OF RECOVERr A 1935 • 1936 HOtLOW SYMBOLS SHOW - 1^37 LOCALITY IS DOUBTFUL Fig. 16.—Location of band recoveries from Canada geese banded at the Jack Miner Bird Sanctuary during the autumn, 1935-1939, and reported recovered in the United States during the season of banding. Fig 18.—Location of band recoveries from Canada geese banded at the Jack Miner Bird Sanctuary during the autumn, 1924 or before, and reported recovered in the United States during 1925-1929. Fig. 17—Location of band recoveries from Canada geese banded at the Jack Miner Bird Sanctuary during the autumn, 1940-1944, and reported recovered in the United States during the season of banding. YEAR Of RECOVERY 1930 '"' HOLLOW SYMBOLS SHOW jlll LOCALITY IS DOUBTFUL 1934 Fig. 19.—Location of band recoveries from Canada geese banded at the Jack Miner Bird Sanctuary during the autumn, 1929 or before, and reported recovered in the United States during 1930-1934. March, 1950 Hanson & Smith: Canada Geese 91 rEAR or RECOVERY 4 )935 '936 HOLLOW SYMBOLS SHOW LOCAUTY IS DOUBTFUL1937 I93B 1939 Fig. 20.—Location of band recoveries from Canada geese banded at the Jack Miner Bird Sanctuary during the autumn, 1934 or before, and reported recovered in the United States during 1935-1939. Bay Company, who was stationed a num- ber of years at Port Harrison, recently stated (personal conversation, August, 1949) that about 75 per cent of the geese killed by the Eskimos in that area were moulting geese that came into the area in early June from the south and not birds that nested locally. His observations thus help to substantiate a relationship that could only be surmised from band re- coveries. The Eskimos in the vicinity of Port Harrison, finding other kinds of game less easily obtainable in summer, turn to the inland lakes, where apparently they secure a plentiful supph- of flightless geese. Sixteen recoveries of Canada geese banded at Horseshoe Lake, fig. 30, and an important percentage of the total re- coveries of geese banded at the Miner Sanctuary in the autumn, fig. 7, have been made in the Port Harrison district. One or more of a number of possibilities may explain these inconsistencies in the re- covery pattern : flights by small groups of Mississippi flyway geese across James and Hudson bays in late summer; north- ward movements by geese of the South- east population along the east coast of the bays for the purpose of feeding on berries; actual intermingling of birds from the different flyways. Trapping at the Miner Sanctuary has shown that some of the Horseshoe Lake geese stop at the Sanc- tuary in the spring, along with the flight of South Atlantic geese. Of 33 Canada geese trapped and banded at Horseshoe Lake and retrapped at the Miner Sanc- tuary, 1943-1945, 11 were retrapped in the spring. The disposition of some Horse- shoe Lake geese to follow the Soutli Atlan- tic geese to the east coast of Hudson Bay would not be surprising. Recovery of F-marked (autumn banded) birds in the Port Harrison district might be partially explained by the banding of South Atlan- tic geese at the Miner Sanctuary in the autumn. A certain amount of overlap in migration routes, with the resultant intermixing at the Miner Sanctuary of South Atlantic geese with Southeast and Mississippi flyway birds, is no less to be TEAR OF RECOVERY • 1940 • 1941 HOLLOW STMBOLS SHOW 943 LOCALITT 1$ DOUBTFUL 1944 Fig. 21.—Location of band recoveries from Canada geese banded at the Jack Miner Bird Sanctuary during the autumn, 1939 or before, and reported recovered in the United States during 1940-1944. 92 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 expected in the autumn than in the spring of the year. West Coast Muskeg Types Aerial reconnaissance flights in the region west of James Bay and south of Hudson Bay, fig. 4, revealed that the mus- keg in the breeding range of Mississippi Valley Canada geese differs considerably in various sectors in the proportions of tim- ber and water it supports. For the sake of convenience the muskeg can be divided into five main types. It must be remem- bered, however, that gradations between all types exist. Type 1. M^ell-timbered muskeg, with only a few ponds or small, widely scattered lakes, fig. 22. Type 2. Open muskeg, with treeless or lightly timbered areas of stunted tama- rack, alternating with small blocks or ex- tensive stands of black spruce, fig. 23. Type 3. Lake-land viuskeg, relatively well-drained areas, more or less timbered, but notable for the numbers of large, widely scattered lakes without islands, fig. 24. Type 4. Pothole muskeg, characterized by a myriad of ponds and small lakes, principally from 5 to 30 acres in size and usually possessing one or more islands. These water areas are often so closely grouped that only small patches or narrow strips of land separate one from the other, figs. 25 and 26. Type 5. "Smallpox" muskeg, that is, muskeg in which sphagnum predominates, the country being more or less a continuous sphagnum bog or series of small bogs in the late stages of filling in so that it can scarcely be classified as land or water, figs. 27 and 28. Fairly extensive areas of this kind occur throughout the Paleozoic Basin and in smaller patches within most areas of the above four types of muskeg. Aerial observations revealed that the Fig. 22.—Type 1 or well-timbered muskeg. The muslceg lying adjacent to the southern half of the west coast of James Bay is fairly well wooded with black spruce and tamarack. Alter- nating with the wooded tracts are extensive areas covered with a heavy growth of willow. Ponds and lakes are relatively few in number in this area. March, 1950 Hanson- & Smith: Canada Geese 93 Fig. 23.—Type 2 or open muskeg. The dark bands across the lower half of this illustration represent stands of black spruce; the lighter colored trees are tamaracks. The spruces are con- fined mainly to better drained sites and to hummocks of mosses and lichens. The tamarack occurs both as light stands on the better drained sites and as scattered, stunted individuals on open sedge areas. In this type of muskeg, the treeless or lightly timbered areas of stunted tamarack alternate with small blocks or extensive stands of black spruce. Fig. 24.—Type 3 or lake-land muskeg. Shown here is an area just north of the Albany River (flight I, fig. 4) about 45 miles inland from the coast of James Bay, 94 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Fig. 25.— lype 4 or pothole muskeg. This photograph was taken a few miles north of the Albany River on flight IV, fig. 4. Fig. 26.—Type 4 or pothole muskeg about 40 miles north of the Moose River and about 30 miles inland from the shore of James Bay. I March, 1950 Hanson & Smith: Canada Geese 95 muskeg for the first 20 to 25 miles inland from the coast of James Bay in the Albany- River country, fig. 4, I and III, exclusive of coastal marshes, is chiefly type 1. The muskeg for the next 35 to 40 miles, or about to the longitude of Fishing Creek Island in the Albany River, is characteris- tically type 4. Alost prevalent in the country between Ogoki and points 50 to 60 miles inland from James Bay, fig. 4, I, are types 2 and 3. From Ogoki, on the Albany River, to a point northwest on the Attawapiskat River, fig. 4, II, muskeg types 2 and 3 characterize the countr\. From this point on the Attawapiskat River to Fort Albany, fig. 4, III, the kinds and the distribution of the muskeg observed are similar, but in reverse se- quence to those seen on flight I, fig. 4. On flight IV, fig. 4, between the Albany River and Weenusk, the follow- ing sequence of muskeg types was found to prevail. Type 4 is dominant between the Albany River and the Atikameg River, wiiich lies 15 to 30 miles north of the Albany; types 2 and 3 most of the way between the Atikameg and Kapiskau rivers, north to the Attawapiskat River, and for an additional 10 to 15 miles beyond. Midway between the Attawapis- kat and Ekwan rivers the muskeg varies between types 4 and 5. Near the Ekwan River, the country appears to be better drained and timbered, and the muskeg of type 1. From the Kkwan Ri\er north- ward, muskeg types 2 and 3 again prevail, but near the Sutton River, which enters Hudson Hay from the southwest at a point 64 miles west of Cape Henrietta Maria, the muskeg is poorly drained and well supplied with lakes of all sizes. From the Sutton River country to Wee- nusk the density of the stands of black spruce decreases and the amount of (Ua- doniu lichen as ground cover steadily in- creases ; in other respects the muskeg ob- served in this part of the flight seems to be either t\"pe 3 or type 5. 1-ig. 27.— lype •> or "smallpox" muskeg, aliout 15 miles nortli ut ihc Aitav\.i|.i>kac Klvcr (flight IV, fig. 4). 96 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Fig. 28.—Type 5 or "smallpox" muskeg, a vertical aerial view. Small bogs, such as these, in various stages of filling in with sphagnum moss, occupy considerable areas of the muskeg country west of James Bay. They contribute little or nothing to waterfowl production. Between Weenusk and Fort Severn, fig. 4, VIII, muskeg types 4 and 5 are most common. However, the lakes in this section of the muskeg, some of which are large, do not appear to offer optimum habitat for nesting pairs of geese, as most of them lack islands. The country be- tween Fort Severn and York Factory, fig. 4, VII, appears to be on the whole rela- tively poor breeding range. In general, the muskeg alternates chiefly between types 3, 4, and 5. West Coast Production Centers On the aerial flights outlined in fig. 4, approximately 217 C anada geese, adults and goslings combined, were observed. From these sight observations, from band recovery data given in table 3, the distri- bution of the various muskeg types, their relation to the configuration of the streams and rivers, and the literature, the exist- ence and location of major production centers, rather than continuous nesting areas, have been deduced. Most of these areas are between two adjacent or con- verging rivers, similar to the river shown in fig. 29, but in type 4 muskeg. Most band recoveries and sight observa- tions of geese can be correlated with the distribution of pothole muskeg, type 4. In the majority of areas in which geese were observed, water areas occupied at least 25 per cent of the surface. This muskeg type occupies slight but extensive depressions or troughs in the Paleozoic March, 1950 Hanson & Smith: Canada Geese 97 Basin, which probably either ( 1 ) origi- nated as depressions in the surface of the glacial drift that mantles the region or (2) developed in connection with deposits accumulated during the uplift of the region, probably at an irregular rate, following its submergence during glacia- tion. These basins now serve as origins of many small streams, while the larger rivers, in seeking the lowest ground when cutting their channels, have tended to converge toward each other in the region of these basins. Consequently, the present-day configuration of the drainage pattern is a clue to the location of pothole muskeg and in turn of production centers for Canada geese. Available information indicates the following production centers, iig. 30, for the Canada geese that use the Mississippi flyway : Production Center A. Between the Al- bany and Attawapiskat rivers in the re- gion of Ogoki and Martin Fall, about 200 miles from the coast of James Bay. Barnston's early report (Richardson 1851 ) and band recoveries point to the presence of this production center, although it seems to be a relatively unimportant one. The localities mentioned above are just within the western limits of the Paleozoic Basin, figs. 2 and 8. In 1947, the 16 hunters in the Ogoki Indian band were questioned regarding the presence of breeding pairs within their trapping terri- tories. Only a few of the hunters had knowledge of Canada geese nesting in the general region north of Ogoki, and they agreed that breeding pairs were scarce in that sector. A single goose v.as sighted in this area on flight II, Hg. 4. Production Center B. Between the Atik- ameg and Albany rivers, from a distance of about 25 miles inland from the coast of James Bay westward to about longitude 82° 50' or the longitude of Fishing Creek Island in the Albany River. Although the country between the Albany River and the Stooping River, a tributary to the south, was not flown over directly, as much of it as could be seen from the plane appeared to be similiar to the coun- try' between the Albany and the Atika- meg and equalU' attractive to nesting geese; probably it should be included as part of the production center. The area m .m^ Fig. 29.—The Attawapiskat River, at a point 30 miles inland from the coast of James Bay. Most of the muskeg shown in this photograph is classified as type 2 or open. Illinois Natural History Survey Bulletin Vol. 25, Art. 3 H U D S N BAY LEGEND RECOVERIES (NOT LISTED IN TABLE 3) OUTER LIMIT OF MAIN BREEDING RANGE PRODUCTION CENTERS SCALE or MILES ^ Fig. 30.—Location of production centers, limits of the main range of the Mississippi Valley geese, and located recoveries in Canada, 1941-1947, of Canada geese banded at the Horse- shoe Lake Game Refuge. Within the main breeding range 217 band recoveries have been made. (Not shovpn are one recovery from Warren, Manitoba, and one from McLean, Saskatchewan.) March, 1950 Hanson & Smith: Canada Geese 99 Fig. 31.—View from the west coast of James Bay about 38 miles south of Lake River. The siuuous tracts of spruce occupy old beach ridges near the coast ; the intervening areas are marsh. Many bands were recovered from Canada geese near this part of the coast. observed, which is characteristically type 4 or pothole muskeg, contained the only geese seen on the east and west flights between Fort Albany and Ogoki. On the return flight, III, tig. 4, 55 adults and 19 goslings were observed. These observa- tions substantiate the location of produc- tion center B up the Albany River, in- dicated earlier by band recoveries, table 3. Field observations and information ob- tained from Indian hunters indicate that few if any geese nest within 10 miles of the shore of James Bay. The Indians report that very few geese breed in the muskeg close to the bay. Most band re- coveries, table 3, from the 9-mile coastal zone probably represent migrating geese shot early in the spring, or wandering, nonbreeding geese. Despite the fact that some of the In- dians from the coastal posts trap and hunt far inland, they have made only a few recoveries of goose bands in the Albany Ri\er district more than 60 miles west of James Bay. Substantiating our own find- ing in the Albany River district, the In- dians report that most of the geese breed within 70 miles of the coast, or not much farther west than 30 miles below the juncture of the Albany and Chipie rivers. Production Center C. Between the Attawapiskat and Ekwan rivers at a distance of between 40 and 50 miles in- land from the coast of James Bay. This area was flown over on northward flight IV, fig. 4, from the Albany River to Weenusk. Band recoveries and aerial observations indicate that this area is a relatively unimportant production center. While its extent east and west can only be surmised from band recoveries, aerial observations indicate that its north and south axis is short, approximately 12 miles. Taken as a whole, the potholes and lakes between the Attawapiskat and Ekwan rivers are in a much more ad- 100 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 vanced state of filling in than are those between the Atikameg and Stooping rivers and few contain islands. Consequently, they are less attractive to nesting geese. On flight IV, fig. 4, three geese were ob- served in this production center, and northwest of this center, about 11 miles south of the Sutton River, a single goose was noted. Production Center D. Between the At- tawapiskat and Ekwan rivers, from 90 to 100 miles inland from the coast of James Bay. A production center in this area is suggested by three recoveries, table 3, and some convergence by the two rivers mentioned, as well as by the drainage pat- tern of the small streams in this area. Production Center E. South of the barren grounds of Cape Henrietta Maria; from about the latitude of Lake River south to the Swan River and at indeter- minate distances inland from the coast of James Bay. The large numbers of re- coveries made along the coast in this area, fig. 31, and the multitude of small, short rivers that drain inland areas in this sec- tor suggest that the production center may lie within 15 miles of the James Bay coast. Perhaps indicative of the approxi- mate location of this center is the Kinu- sheo River, which originates in this region and flows to the northwest to empty into Hudson Bay. When the latest 8-miles to 1-inch maps, based upon high altitude photography carried out in 1947, are com- pleted, the limits of this center will be more easily ascertained. Production Center F. Between the Winisk River and the Fawn River, at a point about 100 miles inland from the coast of Hudson Bay. In this sector the Winisk River and the Fawn River, the latter a tributary of the Severn River, bow sharply toward each other. Between these rivers a dendritic drainage pattern with a number of poorly defined lakes is shown on an 8-miles to 1-inch Canadian topo- graphic map. At Weenusk, where there are some fairly suitable nesting lakes close to the coast of Hudson Bay, the Indians report that they shoot most of their banded geese, table 3, about 150 miles up the Winisk River in the general region outlined above. The winding of this river accounts for the difference in the two mileage figures given for the location of this production center. Map and band recovery data and the size of the kills made by the Weenusk Indians indicate that this production center is second in importance only to the one between the Atikameg and the Albany or Stooping rivers. There is probably some scattered nest- ing over a large area south of Weenusk. On flight IV, fig. 4, two flocks, one of 21 geese and another of 6 with goslings, were sighted about 33 miles south of the Winisk River at a point about 25 miles from the coast of Hudson Bay. On flight VIII, fig. 4, between Fort Severn and Weenusk, 15 Canada geese were observed from the air. However, the lakes flown over on flight VIII did not appear to offer op- timum habitat for nesting pairs, as they generally lacked islands. The Weenusk Indians say that they find breeding pairs nesting closer to the coast in early, mild springs than in late, cold springs. Production Center G. Severn River country. One or perhaps several pro- duction centers, poorly defined in either case, may lie in the Severn River country. The configuration of the river and its tributaries and two band recoveries sug- gest that a production center may be found somewhere between 50 and 90 miles up this river. William Glennie, a post manager for the Hudson's Bay Company, told the senior author in 1947 that he had seen fresh goose eggs that were taken from nests found in the upper portions of the Severn River watershed, between Windi- go and Big Trout Lake, localities that lie just west of the Paleozoic Basin, rough'y between latitudes 52° 30' and 54°, but stated further that the greatest numbers of Canada geese were found along the lower portions of the Severn River. On flight VII, from York Factory to Fort Severn, a distance of about 145 miles, 28 Canada geese were observed, a number that is indicative of a low population density in this section of the Paleozoic Basin. Observations and aerial photos reveal that the habitat in this area is of relatively poor quality. Many of the water areas are in the late stages of filling in and the great majority of lakes lack islands. Nevertheless, a portion of the muskeg west of Fort Severn probably March, 1950 Hanson & Smith: Canada Geese 101 itik^^iife 1*?^. Fig. 32.—During the spring and summer the muskeg country is very difficult to traverse on foot. In scene pictured here, small spruce are being put down in order to permit crossing between two mats of floating sedge. The photograph was taken in the Lawapiskau River country on July 3, 1947, at which time ice still could be found in many places 15 inches below the surface of the sedge mat. should be included in the general breeding range of the Mississippi flyway population because bands have been reported from this country. The breeding density of Canada geese is probably greater than the small number of band recoveries indicate for this section of the muskeg, fig. 30, be- cause of the preference of the Indians at York Factory for hunting other kinds of geese on the coast of Hudson Ba\ : Richardson's goose (lirunta hutchitisii) and the lesser snow goose (Chen It. Iiyper- borea) , species said to be fat both in the spring and in the autumn, while the Canada goose is reported to be thin and unpalatable when it arrives on the in- terior breeding grounds. Production Center H. Akiiniski Island. From the accounts of the Indians at At- tawapiskat and observations made from the air, nesting on Akimiski Island is found chiefly in the central portion close to the south coast. In this area many suit- able lakes were seen and 61 geese were observed on the 1947 flight. Nest Sites Although there are a number of fairly well-defined centers of production where most of the geese nest and rear their young, aerial flights in 1947, fig. 4, sub- stantiated the information gathered earlier from the Indians that the breeding pairs are scattered within these centers; there is seldom more than one pair on a given lake. Further evidence pointing to scat- tered nesting was gained by the senior author in 1947 when traversing the mus- keg on foot. A few penetrations of the muskeg were made at points 15 and 25 102 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Fig. 33.^Typical type 4 muskeg lake with small islands. According to native Indians, small islands in lakes of this kind offer preferred nesting sites to Canada geese. Fig. 3-t.—Vertical view of type 4 or pothole muskeg. Extending outward from most stands of trees is a floating mat of sedge partially supported by sphagnum moss. March, 1950 Hanson & Smith: Canada Geese 103 miles up the Lawapiskau River, reported- ly goose-nesting country, and at two points 40 miles up the Albany River, which the aerial flights a few days before had revealed as production centers. Dur- ing these walks, only one pair of geese was observed, but the faint trails of several broods were found, revealing where geese had moved from one small muskeg lake to another. Unfortunately, because the breeding pairs were scattered and the nesting habitat was highly inaccessible, both from the standpoint of getting a canoe within walking distance of a production center and of actually traversing it on foot, fig. 32, no nests were located in 1947. Ac- cording to the Indians, small islands in lakes and ponds offer preferred nesting sites, figs. 3i and 34, but, where no is- lands are present, nearly any location close to the water's edge is suitable. An impression gained from the aerial survey is that small lakes of 5 to 30 acres in size and possessing one or more small islands are the type preferred by nesting pairs. In the western United States, Canada geese have been found by wildlife workers to concentrate in favored sections of a marsh or breeding range, such as partic- ular islands in lakes and reservoirs. As a result of such colonial-type nesting, young broods of several pairs frequently combine into a large rearing brood, a single pair eventually taking charge of this brood. The fact that only families of normal size have been observed at Horse- shoe Lake, or have been reported by Jack Miner at Kingsville (see section on "Pro- ductivity"), suggests that scattered nesting is the rule in the muskeg west of James Bay ; the assumption is that nesting pairs are so spaced that contacts between broods are infrequent and combination does not take place to an important degree. Information corroborating this view- point was reported by R. M. Duncan and A. H. Michell. Both of these men have spent many years as post managers on the east and west coasts of James Bay. They report that the autumn migration of Canada geese along the west coast is primarily that of small family flocks, observations which are in agreement with those made by the authors. On the east coast, according to Duncan and Michell, flocks of from 20 to 40 or more geese generally comprise the autumn flight. However, the presence of large flocks in the autumn along this coast of James Bay is not surprising since areas of favorable habitat are more limited there and the densit)- of nesting pairs is relatively high, particularly on the Belcher and Twin islands. Donald F. Coates and Donald B. Coombs, as cited earlier (personal com- munication, 1947), found 6 adults and 21 goslings together on \Veston Island. From these reports it would .seem that in the Hudson and James bay region, as in western United States, crowding of the nesting pairs is a factor likel\- to induce the combining of broods. MIGRATION The beautiful and often spectacular flights of the Canada goose have prob- ably held a greater fascination for more people than the flights of any of our other native birds. Some persons think of geese in flight as special creations, living en- viable and unfettered lives. Other per- sons thrill to the sight of migrating geese as an object of sport. To the Canadian Indian trapping in the "bush," the first flocks of geese in early spring afford a welcome opportunity for a change of diet from bannock, beans, and dried or salted meats. In years when fur and game animals are at low points of their cycles, and consequently food stocks are close to depletion, the arrival of geese may mean relief from near starvation. Autumn Migration Routes Our data on the movements of the Canada goose in the Hudson-James bay area are based on information received from the Indians and white residents and on personal observations. Band recov- eries have been the principal source of in- formation relating to autumn migration movements of Canada geese in the United States, figs. 13—21, but these recoveries do not furnish a complete picture of the migration routes. Naturally, most re- coveries are from localities where hunters as well as geese congregate, generally in the vicinity of favorite waterfowl rest lakes or feeding areas where the flocks linger before continuing south. Wooded 104 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 or hilly country and waterless prairies are usually flown over nonstop by migrating geese. Consequently, very few bands have been recovered from country of this na- ture, even though large numbers of geese pass overhead each autumn and spring. Fig. 35.—The tundra of the Cape Henrietta Maria area as seen from the air. Fig. 36.—The mouth of the Moose River and a portion of Ship Sands Island. The extensive marshes shown in this scene are heavily used by blue and snow geese and to a lesser extent by Canada geese in the autumn. March, 1950 Hanson- & Smith: Canada Geese 105 In Canada.—Before the southward migration from the breeding grounds takes place, a rather complex series of local flights occurs. About August 15, short- ly after the young birds are on the wing, a movement begins to the coasts of Hudson and James bays. This is not a mass flight but a movement of family groups and small flocks from some of the produc- tion centers near the coast. The geese that have nested adjacent to the south- west coast of Hudson Bay fly north to the coast and then almost due east to Cape Henrietta Maria, figs. 2, 30 and 35 ; those that nested adjacent to the west coast of James Bay, north of the Ekwan River, fly east to the coast and then north to Cape Henrietta Maria. This cape is an isolated area of tundra attractive to the geese at this season because of the abundance there of blueberries {J'accir.iiitn sp.), billberries (1 acchiiurn uli'jlnosum \ , dwarf rasp- berries {Riibus (irctirus), and crowberries (Empelrum n'tijrtitn). It is of interest to note here that flights to the sea coasts for the purpose of feeding on berries and other foods have been reported for other Canada goose populations in the north country [Newfoundland ( Howley 1884); north- ern Unga\a (Bent 1925, quoting Lucien M. Turner) ; and Labrador (Austin 1932)]. The geese that concentrate on the tun- dra of Cape Henrietta Maria remain there for varying periods before fl\ing south. The length of time the geese re- main in this region depends to a large degree upon the success of the berry crop, but probably all geese leave the cape by the latter part of September. At least half of the "cape geese," as they fly south down the west coast of James Baj, stop at Akimiski Island, where they concen- trate on the wide flat marsh on the north side, a favorite feeding area. According to A. H. Michell of the Hudson's Bay Company, this Hijjht usually takes place about September 15. Most of the gee;e nesting south of the Ekwan River remain in the interior, al- though a few of them fly to James Ra\ , where they congregate in moderate-sized flocks in the coastal marshes about the river mouths, in country similiar to that shown in fig. 36; others continue to the marshes of Akimiski Island. Apparently, many of the geese that feed on Akimiski Island fly directly to the Jack Miner Sanctuary as soon as they lea\e the island. The Indians at Fort Alban\' claim that since Jack Miner started band- ing geese at his sanctuary they ha\e killed only a few in the autumn. A number of bands from geese banded at Horseshoe Lake have been recovered in summer from the Belcher Islands and the east coast of Hudson Ba\ in the region of Port Harrison, fig. 30. ^V^hether these bands have been recovered from South Atlantic geese that strayed from their normal flyway and were banded at Horseshoe Lake, whether they were re- covered from Mississippi flyway geese that strayed east of their normal flyway on their spring migration, or whether they were recovered from Mississippi flyway geese that nested west of James Bay and then struck out across the bays can be only conjectured on the basis of available data. In any case, many geese that are on the east coast of Hudson Bay in autumn migrate southward along the coast to the south end of James Bay, where they converge with the groups that have flown south along the west coast of James Bay. Because band recoveries suggest a northward movement along the east coast of James and Hudson bays in the early autumn by geese that have nested inland from the south coast of James Bay, we believe that the final southward flights along the east coast of both bays may consist of at least some geese from three different populations. Geese of the South Atlantic population that nested along the east coast and on neighboring islands (Belcher and others) make up most of the flight ; geese of the Southeast popula- tion probably are second in numbers ; while individuals of the Mississippi V^alle\' population are least numerous, fig. 7. At points near the south end of James Bay, the South Atlantic geese split away from the Mississippi \'a!ley and South- east populations; portions of only the last two populations migrate through the Kingsville region in the autumn. The apparent mechanism of the splitting off of the southward flights along the east coast of James Bay into their various components, fig. 37, has been deduced 106 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Fig. 37.—Probable migration routes taken by various populations of Canada geese at the southeast end of James Bay. The first splitting away from the combined flocks that migrate southward down the east coast of James Bay occurs somewhere along the northeast shore of Rupert Bay. A second splitting away occurs at or near the feeding grounds bordering Cabbage Willows Bay, the Mississippi Valley geese flying southward, the South Atlantic geese toward the southeast. .March, 1950 Hanson- & Smith: Canada Geese 107 from a description of goose flights in the Rupert House area given to us by A. H. Michell, post manager at Rupert House. The first splitting off of the combined autumn flights evidently occurs some- where along the northeast shore of Ru- pert Bay, fig. 37. Some of the birds follow the northeast shore of Rupert Bay to the coastal marshes near the mouth of the Rupert River, where they congregate and feed ; then they leave the James Bay region and fly southeast. Other flights cross Rupert Bay and feed in the marshes in the vicinity of Cabbage Willows Bay. At this point a second split occurs; some of the geese fly southeast, while the remain- der follow a natural pass along a small stream and a series of muskeg lakes across the neck of the Ministikawatin Peninsula. These birds continue on to Hannah Bay, where they find final feeding grounds be- fore departing from the James Bay region. The geese that have remained in the muskeg west of James Bay, instead of flying to the coastal marshes, migrate south on a broad front, crossing into upper Michigan, Wisconsin, and eastern Minne- sota, fig. 13. Probably they comprise the majority of the birds in the Mississippi Valley population, figs. 13—21. In the United States.—Band re- coveries indicate that the flights of Canada geese that enter the United States by way of upper Michigan, Wisconsin, and east- ern Minnesota constitute the bulk of the Mississippi \'alley population figs. 13-21. The flocks that migrate through Wiscon- sin in the autumn adhere principally to the eastern half of the state. Alany of the flocks follow the west shore of Lake Michigan. Other flocks favor one of two other routes : ( 1 ) the \alley of the Wisconsin River; (2) from Green Bay south to Lake AV'innebago, the flights probably splitting south of Lake Winne- bago, one sector going to the Lake Geneva area and the other following the Rock River. According to Zimmerman (1943), the greatest concentration of Canada geese in Wisconsin during the autumn migra- tion occurs in .Adams, Columbia, Fond-du- Lac, Sauk, Walworth, and AVaushara counties; the Arlington prairie in Colum- bia County and the Rock and Big Foot prairies in Walworth County attract the greatest numbers. At the peak of the flight in 1941, about November 15, it was estimated that 15,000 to 20,000 geese were using Lake Wisconsin (Zimmerman 1942). Five thousand of these birds fed in the cornfields in the vicinity of Sump- ter, Sauk County (Zimmerman 1942). Appreciable numbers of Canada geese follow the west shore of Lake Michigan south, according to A. B. McDonald of AVadsworth, Illinois, who reported to Frank C. Bellrose of the Illinois Natural History Survey that each year flocks of Canada geese follow the shore line as far south as Zion, Illinois, at which point they leave the lake and fly southwestward. The exact route taken each year is said to remain identical. The Canada geese entering the United States from the Miner Sanctuary by way of southeastern Michigan or northwestern Ohio constitute a part of the Mississippi \'allcy population. Reco\eries of geese banded in the autumn at the Miner Sanc- tuary show that this segment of the Mis- sissippi rtyway population migrates almost straight southwest to the Ohio or lower Wabash rivers, stopping en route in con- siderable numbers at Lake St. Mary or Grand Reservoir, a 17,500-acre impound- ment lying in Mercer and Auglaize counties, western Ohio. On leaving Lake St. Mary this group seeminglj' flies di- rectly to the Ohio Ri\er valley, which it follows to Horseshoe Lake. Another group of geese appears to migrate across lower Michigan from Saginaw Bay to the counties in the south- western portion of the state. Some of these geese winter in the vicinity of the W. K. Kellogg Bird Sanctuary near Gull Lake, Kalamazoo County, and along the lower Kalamazoo River. The majority eventually continue southward, crossing north-central Indiana to the Wabash River bottoms; some of them join geese that have migrated south along the east shore of Lake Michigan and then fly either straight south to the Wabash and Ohio river bottoms or in smaller numbers fly southwestward directly to Horseshoe Lake. Much of the Canada goose flight enter- ing Illinois from Wisconsin in all likeli- hood traverses the length of Illinois on a fairly broad front, but band recoveries 108 Illinois Natural History Survey Bulletin Vol. 25, Art. i X s B 3 « O o .a o O j: §3 Hi March, 1950 Hanson ,^ Smith: Canada Geese 109 suggest that important numbers of birds follow the Illinois River to its juncture with the Mississippi River. At that point the Illinois River flight may be augmented by flocks (relatively few in number) that follow the Mississippi River southward toward Horseshoe Lake. Because there has not been sufficient banding of Canada geese in southern parts of the Mississippi flyway, the flight lanes of Canada geese wintering on the lower Mississippi River, from Tennessee to Louisiana, are less apparent than the routes taken by flocks wintering farther north. Recoveries of geese banded at Kingsville, Ontario, in the autumn, figs. 1-1—21, suggest that many of the flocks migrate down the lower Ohio River valley to the Tennessee River, which they follow south instead of continuing on to Horseshoe Lake. Presumably, at a num- ber of points these flocks later leave the Tennessee River and cross over to the lower Mississippi River. Additional data indicating that con- siderable numbers of geese by-pass Horse- shoe Lake to the east via the Tennessee River are found from band recoveries of geese raised at Seney National Wildlife Refuge in the northern peninsula of Michigan, fig. 39K. Of the total number of band recoveries made, the number re- ported from Arkansas was second only to the number reported from Michigan (Johnson 1947). These recoveries were made during the same period that hea\y kills were occurring at Horseshoe Lake. It is, of course, assumed that the migrant birds from the Seney Refuge joined other wild flocks from the north or at least used the traditional paths of migration. Recently Earl L. Atwood, manager of the Kentucky Woodlands National Wild- life Refuge, informed the senior author (personal communication, December, 1947) that the Tennessee River valley is a traditional flyway for Canada geese. There is no evidence, either from obser- vation or from band recoveries, to indicate that there is an important turnover in the flock using the Horseshoe Lake area in the autumn. According to our records, only one goose banded at the refuge has been taken an appreciable distance south of it the same season as banded, fig. 13 and table 4. If a turnover in the flock occurs, it must be early in the autumn before many geese ha\e been banded. There is reason to believe that the Horseshoe Lake Refuge has acted as a "bottleneck" in that each year it has at- tracted increasing numbers of geese that previously have wintered along the lower Mississippi River. Few of these geese, having entered the refuge, woXild be ex- pected to continue migration later in the season, ex'cept under pressure of extreme weather. Hence, they would augment the concentration surviving from previous years as well as contribute to the kill. The theory that the refuge acts as a "bottleneck" assumes that ingress of new birds from other areas exceeds the egress of old flock members. The decoying effect of a large concentration, abundant food, and a roost lake would seem the basis for a differential in favor of ingress. We do not yet have satisfactory data on tlie migration routes of Canada geese wintering in western Louisiana. Re- covery records of geese banded at the Miner Sanctuary and at Horseshoe Lake indicate that the migration routes of the Mississippi Valley population do not lie far west of the Mississippi River. Hence, the flocks that migrate through central or western Minnesota, Iowa, Missouri, and Arkansas, and those (possibly the same) that winter in western Louisiana, constitute part of a distinct population, the Eastern Prairie, but scattered band recoveries of geese banded at Horseshoe Lake and taken in Manitoba, South Dakota, western Minnesota, Louisiana, and eastern Texas are evidence that there is some exchange of birds between the Mississippi Valley population and the Eastern Prairie population. Spring Migration Routes There are too few spring band re- coveries in the United States to depict ac- curately the northward migration routes of Mississippi flyway geese. Judged by trap records from the Miner Sanctuary, the spring movement is more directly northward and somewhat west of the autumn migration routes. Each spring in early March, a marked increase is noted in the numbers of Canada geese at Horseshoe Lake and at Hovey Lake, Posey County, Indiana. The latter area 110 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 harbors few geese in the autumn, but is host to large concentrations after the middle of February. It is conceivable that the late winter concentration may consist of geese of the Southeast popula- tion, which may take a more westerly route in their northward than in their southward migrations. Other important late winter or spring concentration points are the drainage districts near Putnam, Illinois (spring 1946 and 1947), the Horicon National Wildlife Refuge (Hop- kins 1947), bottomlands of the Bark River in Wisconsin, farm lands at the south end of Lake Oshkosh in eastern Wisconsin, and Gull Lake in southwest- ern Michigan. As most spring band recoveries are from the remains of geese shot the previous autumn, no differentiation is made be- tween autumn and spring recoveries in figs. 13-21. Apparently, after feeding in the rich farm lands along the migration routes in the United States and southern Canada, the flocks fly almost directly to the breed- ing grounds. Time and Rate of Migrations The autumn migration of geese win- tering at Horseshoe Lake is spread out over at least a 3-month period, the earliest migrants leaving James Bay in the fore- part of September and the last reaching Horseshoe Lake in December, the exact dates depending on the severity of the weather. Migration records from federal refuges and Horseshoe Lake, table 5, suggest that the outward movement of geese from the breeding grounds may be compared with a segment of the concentric waves pro- duced by an object striking the surface of a body of water ; the earliest flocks or migratory waves travel the greatest dis- tances in the shortest periods of time and reach their wintering grounds in the far south before many other flocks hsive left the north country. First arrivals are noted at Horseshoe Lake and at federal refuges farther south as early as or earlier than they are recorded at ref- uges farther north. A similar picture has been found to be true for areas lying only short distances apart. Leopold & Jones (1947) reported that in 5 out of 6 years flocks of Canada geese were recorded near Madison, in Dane County, Wisconsin, 2 to 27 days before they were observed about 40 miles to the north- west, near the Wisconsin River, in Sauk County. Table 5.—Dates of first recorded autumn arrivals of Canada geese at federal refuges and at Horseshoe Lake, Illinois, 1938-1941. Refuge March. 1950 60,000-1 Haxsox & Smith: Canada Geese 111 50,000 40,000 13 30,000- ? 20,000 HORSESHOE LAKE, ILLINOIS 1941-42 1942-43 1943-44 1944-45 1945-46 ROCK-WALWORTH COUNTIES, WISCONSIN iiii l ii 1942-43 NUMBER AT INVENTORY 20 25 30 SEPTEMBER 10 15 20 OCTOBER 15 17 25 28 JANUARY Fig. 38.—Build-up of the Canada goose flock at Horseshoe Lake during the autumn and winter of the years 19+1-1946. .\lso shown is the build-up of the Canada goose tloclc in Rock and Walworth counties in the autumn and winter of 1942-43 (from Zimmerman 1943). The build-up of autumn concentra- tions at Horseshoe Lake is shown in fig. 38. As the majority of the geese win- tering at this refuge arrive before the bulk of the kill has been made farther north, probably the flocks that leave the breeding grounds later, and winter farther north, contribute most to the kill in areas north of the refuge. Spring migration movements appear to be more leisurely than the flight south, but this impression may be created by flocks of nonbreeding adult or yearling geese that are under no stimulus to reach the breeding grounds at an early date. VVe have seen several hundred geese in the vicinity of Lake Wisconsin throughout the first week in May. and Hopkins (1947) states that the last flock in the Horicon Marsh area in 1947 remained until May 8. On the other hand, the arrival of the first flocks in the James Bay region is quite punctual, generally be- tween April 15 and 25, which is the time of the goosemoon, "nisku pesim," of the Cree Indians. In most years, the earliest flocks arrive on the breeding grounds 2 to 3 weeks before the break- up of the major rivers, table 6. George MacCloud, a lifelong resident of the James Bay area, reported to the senior author that a second flight of Canada geese generally takes place about Tune 10. These late geese are said to be in large flocks, whereas most of those that arrive earlier are paired. He thought that the late arrivals were largely young of the previous year. Although we have been in the bay area during June, we are unable to confirm, by personal observa- tion, the "flight of stragglers." However long the northbound Canada 112 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Table 6.—First arrival or first kill of Canada geese at Fort Albany, Ontario, and date of breakup of the Albany River.* Year March, 1950 Hanson ,:x Smith: Canada tjEssH IIJ vides acceptable roosting sites. AV'hcn the local food supph is exhausted or co\ercd with snow, or when feeding is curtailed, as at the Miner Sanctuary in early Decem- ber, the geese in the northern sectors of the Hyway migrate farther south. -WINTERING -STOPOVER - JACK MUTER MIGRATORY BIRD SAIICTUARY - ILLINOIS RIVER VALLEY - HORSESHOE LAKE GAME REFUGE - GULL LAKE AND W. K. KELLOGG BIRD SANCTUARY AREA - KALAMAZOO RIVER SWAMPS AND MARSHES - LEIDY LAKE - SAGOJAW BAY - LEELANAU AND BENZIE COUNTY AREA - SENEY NATIONAL WILDLIFE REFUGE - ALPENA SANCTUARY M -- ROCK PRAIRIE STATE REFUGE N — GREENWOOD FARM STATE REFUGE - GRAND RESERVOIR OR LAKE ST. >AARYS - HOVEY LAKE STATE REFUGE - LOWER MISSISSIPPI RIVER - WHITE RP/ER NATIONAL WILDLIFE REFUGE - DELTA NATIONAL WILDLIFE REFUGE - WHITE LAKE AND LACASSINE NATIONAL V/ILDLIFE REFUGE - CALCASIEU LAKE AND GALVESTON BAY AREA AREA AREA Fig. 39.—Location of important concentration areas for Canada geese of the Mississippi Valley population. ' 114 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Each January an inventory of the waterfowl populations wintering in the United States is made by the United States Fish and Wildlife Service and co- operating agencies, assisted by selected private individuals. The January inven- tories have produced useful information, particularly in regard to population trends but, because these inventories are taken over a limited period of time (4 days), in some areas they have been subject to considerable error in past years. For example, the immense coastal marshes of Louisiana, which are notoriously difficult to traverse on the ground, cannot be covered adequately except by plane. Be- cause thorough aerial censuses of the Canada goose population in Louisiana were not made before the winter of 1943—1-4, and because adequate data are lacking for many other parts of the fly- way prior to that winter, we do not con- sider the population data previous to that date to be of sufficient reliability to meet present-day management needs. Even some of the data in table 7, particularly the 194-1—1-5 figures for the populations in Arkansas and on a considerable portion of the lower Mississippi River, may not be of sufficient reliability because complete coverage by aircraft was not possible. For reasons explained in the section "Autumn Migration Routes," we believe that the flocks of western Louisiana prob- ably are not an integral part of the Mis- sissippi Valley population. Nevertheless, they should be considered along with the Mississippi Valley population in order to detect whether major population shifts occur between the flyways in some years and to determine the effect that kills in the upper Mississippi River valley may have .on the western Louisiana populations. A brief survey of the various concentration areas and the populations using them follows. Jack Miner Bird Sanctuary The Jack Miner Bird Sanctuary, lo- cated in the rich farm lands of Essex County, Ontario, figs. 39A and 40, 4 miles from Lake Erie, was one of the first waterfowl refuges established in North America. The history of this ref- uge and of Jack Miner's work with Canada geese has a bearing on discussions Table 7. — Population of Canada geese in the Mississippi River valley, 1943-44 through 1946-47.. Data are from the annual January inventories, except as noted. State or Other Area Seasok 1943-44 1944-45 1945-46 1946-47 Michigan . . Wisconsin. . Minnesota . Ohio Indiana Illinois I Mason County). Horseshoe Lake Iowa Kentucky Mississippi River (Tenn.-Miss. line to White Ca.stle, La.) Arkansas Missouri Louisiana Delta Western Louisiana Tola/ Total, exclusive of western Louisiana 37 ,220 ,350 248 343 9251 ,000' 125 ,720 ,3005 ,0005 ,300 ,000* ,000* ,731 ,731 2,200 4,100 5 750 8001 30,000* 6 2,280 10,000" 10,500 5,440 1,0005 12,0005 79,181 67,181 2,343 4,310 100 985 360' 22,000' 1,200 1,650 5,400 665 }l0,000' 49,013 3,512 5,000 105 1,369 31,649' 1,230 7,540 800 2,370 8,065' '61,640 'Census by Frank C. Bellrose, Illinois Natural History Survey. ' Of the number ol Canada geese in Illinois, about 30,000 were at Horseshoe Lake and 800 at the Union County refuge. ' Census by Paul S. Smith, United States Fish and Wildlife Service, and Harold C. Hanson at January inventory. * Average ol estimates by Robert H. Smith, Paul S. Smith, and Frank C. Bellrose after hunting season. * Census by Robert H. Smith. •Total for Tennessee and Mississippi combined in January inventory. ' Inventory figure lor all of Louisiana. According to Richard H. Griffith, United States Fish and Wildlife Service, 1,500 Canad.i geese were at the Delta National Refuge and 5,440 at the Lacassine and Sabine National Wildlife refuges in' the winter of 1946-47. March, 1950 Haxsox & Smith: Canada Geese 115 in this paper and is also of general in- terest. Jack Miner (1923) built his first pond and set out decoys to attract geese in 1904, but did not lure in a family of geese until 1908. The numbers of geese using the refuge built up slowly in the early years, acre homestead area. About 100 acres are planted to rye and timothy, the re- mainder to corn, which constitutes the only grain fed to the geese. Fields of timothy, which have been cut for seed, are said to make ideal pastures for Canada geese and are heavil\' grazed. Approxi- Fig. 40.—View of the main pond and feeding grounds at the Jack Miner Bird Sanctuary, Kingsville, Ontario. Contact of the geese with human beings is avoided whenever possible. Corn is distributed at night, and visitors remain concealed while observing the geese. and until 1915 the refuge attracted Canada geese only in the spring. In later years the autumn flight equaled the spring flight in size. Efforts at trapping and banding Canada geese did not succeed until 1915, and large-scale bandings were not accomp- lished until nearly 10 years later. Table 2 presents the best available data on the numbers of Canada geese banded at the Jack Miner Bird Sanctuary in the autumn. The Miner homestead, ponds, and feed- ing grounds consist of 17 acres. All feed- ing is done around the ponds, but a few rye and timothy fields are planted as re- treats and sources of food to be used when the geese on the ponds are disturbed. Additional farm land, owned by the Jack Miner Migratory Bird Foundation, Inc., amounting to 400 acres, surrounds the 17- mately 20,000 bushels of ear corn are fed during the autumn and spring seasons; when there is an appreciable local kill the corn is fed more heavily than at other times. By Proclamation and Order in Council of the Provincial Government, no shoot- ing is permitted on an additional 1,600 acres of land neighboring the 400 acres owned by the Miners. Thus, the geese are protected in all directions from the central feeding grounds by a buffer strip about 1 mile deep. A few geese arrive at the refuge by late September. Noticeable increases in num- bers usually occur between October 10 and 15, and peak numbers are reached by about November 10. There is a constant renewal of the population as some individ- uals continue their migration south and others arrive from the north. The bulk 116 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 of tlie autumn flock leaves by late Novem- ber or early December. In some years prior to World War II, as many as 5,000 geese were reported to have remained all winter. Some of these wintering geese from the autumn flight have received "S" marked bands in the spring along with birds that have wintered at Curri- tuck Sound and Lake Mattamuskeet, thereby explaining why some spring bands are subsequently recovered in the Mis- sissippi River valley, fig 12. Illinois Of the areas in Illinois important to migrating and wintering flocks of Canada geese, the two most Important are the Illinois River valley and the Horseshoe Lake Game Refuge in Alexander County at the southern end of the state. Illinois River Valley. — Canada geese have been reported from 23 bottom- land lakes in the Illinois River valley, fig. i9B, but regularly from only seven lakes, table 8. These lakes act chiefly as roost areas ; feeding is done in the cultivated uplands and in some drainage districts. Geese of five of the Illinois concentrations disperse to feed as follows : Beebe Lake geese depend largely on the winter wheat and the corn of Duck Island ; Lake Chau- tauqua and Clear Lake geese feed mainly in the cultivated fields of Mason County ; Crane Lake and Jack Lake birds seek most of their food in or near a 1,000-acre private club 2 miles southeast of Bath. Flocks frequenting Goose Pond and Lake Senachwine (the part formerly known as Swan Lake) have not been observed feed- ing in any particular sector. In general, feeding areas are within 7 miles of a roost lake. Population data for the above areas are summarized in table 8. Horseshoe Lake Game Refuge. — The most important Canada goose winter- ing ground in the Mississippi River valley in recent years, the Horseshoe Lake Game Refuge, with its surrounding area, figs. 1 and 39C, during the period of this study harbored approximately 50 per cent of the goose population of the flyway for varying autumn and winter periods. Be- cause of inadequate food supplies on the refuge, as well as intense hunting pressure in surrounding privately owned fields, the flock fed in most winters over a 15-mile radius. The majority of the geese roosted within the refuge each night, although some flocks resorted to islands and bars in the Mississippi River. The lake, fig. 41, 1,200 acres in size, of an oxbow type common to the bottom- lands in the flood plain of the Mississippi River, in many places is 200 or more yards in width and 4 to 6 feet in depth. A dam maintains fairly stable water levels, but most of the land enclosed by the lake is subject to flooding when the Mississippi or Ohio River reaches high flood stage. Open water surrounds the island except for a Table 8.—Canada goose populations autumns of 1938-1946. three regions of the Illinois River valley, • Yf.ar March, 1950 Hansox & Smiih: Canada Geese 117 Fig. 41.—View of the east arm of Horseshoe Lake. The large open expanses of the lake are favored by the geese for roosting purposes. Fig. 42.—The greater portion of Horseshoe Lake is open water, but the north and south portions have heavy stands of live and dead cypress and tupelo gum trees. A dam maintains fairly stable water levels except when the Mississippi or the Ohio River reaches high flood stage. 118 Illinois Natural History Survey Bulletin Vol. 25, Art. .S Fig. 43.—Aerial view of Canada geese on Horseshoe Lake in November, 1945. The popula- tion of the entire flocli could be counted with a considerable degree of accuracy if suitable aerial photographs were available. small portion at the north end, where the lake is svvamplike and has an irregular stand of tupelo gum and cypress trees, fig. 42. Gums and cypresses border the re- mainder of the lake, and in some places the cypresses extend entirely across the lake. During late years of this study the ref- uge contained about 3,660 acres. The Table 9.—Number of Canada geese using the Horseshoe Lake Game Refuge, 1928-29 through 1946-47. Season' March, 1950 Hanson & Smith: Canada Geese 119 i.-.land has an area of 1,360 acres, of which 1,200 have been farmed in recent years to produce food for the geese. The re- maining acreage supports some of the finest virgin bottomland timber in the state. Of the cultivated portion of the island usually 300 to 400 acres are planted to corn and 700 acres sown to wheat, but these acre- ages have varied considerably from year to year. In the last several years all crop land on the island has been planted in corn. Wheat or corn is sown on the 100 acres of the refuge adjoining the east shore across from the island. Many of the published statements in recent years regarding the size of the Canada goose flock at Horseshoe Lake have not been in agreement. The result uas been confusion in the minds of the public. WTiile a few "census figures" have been based on pure guesswork and are therefore unreliable, many of the dif- ferences in published data may be related to the times of the year the censuses were taken, and whether they included only the number of birds alive on certain dates or the total number arriving at the refuge in any given year. The population data given in table 9 summarize the census figures for several years. Population estimates of the Horseshoe Lake flock since 1939 have been made by staff members of the United States Fish and Wildlife Service and the Illinois Nat- ural History Survey, table 9. These estimates have been made by visually dividing the flocks into blocks, counting the number of geese in the sample blocks when the great bulk of the geese are feeding in the wheat fields on and near the refuge, and then using the sample counts to calculate the total population. The practice in some years has been to make periodic estimates from the time the first geese arrived in late September until peak populations have been reached in late autumn. Since 1944, aerial censuses just before and after the hunting seasons have Fig. 44.—.Aerial view of Burnham Island ami adjacent bars in the Mississippi River, 4 miles west of Horseshoe Lake. Prior to the establishment of the refuge, Canada geese wintered in large numbers on similar bars and islands of the Mississippi River, from Chester to Cairo, Illinois. Geese have made some use of these islands even since the refuge was established. 120 Illinois Natural History Survey Bulletin Vol. 25, Art. .1 been made at Horseshoe Lake and nearby areas, figs. 43 and 44. Population figures for 1941-42 through 1945-46 are shown grapiiically in fig. 38. Michigan In Michigan there are three major con- centration areas and two of minor impor- tance. Kalamazoo River Bottoms and Nearby Lakes.—The Kalamazoo River bottoms and a number of lakes in the southwestern section of the state con- stitute the most important region in Michigan for concentrations of migrant and wintering Canada geese, fig 39. This general area includes three specific con- centration sites. 1. Gull Lake and the W. K. Kellogg Bird Sanctuary and Farms are located in Prairieville and Barry townships in Barry County and in Richland and Ross town- ships in Kalamazoo County, fig. 39D. Gull Lake, with an arta of 3,000 acres, is designated as a rest lake. -Hunting is prohibited on the quarter-mile strip sur- rounding this lake and on the Kellogg tract of 600 acres. The above district lies on an extensive outwash plain and is characterized by small lakes and kettle holes. Some near- by sections are too hilly to be farmed, but hay, corn, and wheat are raised extensively on the less hilly sections. The geese feed in the cultivated upland fields and also they are hunted there. In 1945, the maximum autumn popula- tion in the area was 5,000 birds, and about the same number were present during the peak of the 1946 spring migration. The wintering population usually varies from 1,000 to 2,000, but may be considerably less for several weeks in midwinter. In 1944-45, 500 geese wintered at Gull Lake (Dr. Miles D. Pirnie, then in charge of the sanctuary, personal communication). Normally a majority of the birds leave by mid-January and return again by mid- February. Weather determines their movements; usually a portion of Gull Lake remains open throughout the winter, and waste grain is generally available in the uplands for geese that winter in the area. 2. The Kalamazoo River swamps and marshes, fig. 39£, principally the Potta- wattomie and Ottawa marsh areas, the latter a part of the Swan Creek Wildlife Experiment Station located in Heath, Manlius, and Valley townships, and the Todd Farm Sanctuaiy in Ganges and Clyde townships near the Kalamazoo River, all in Allegan County, are some of the most important concentration grounds for Canada geese in Michigan. Each site differs somewhat from the others and there is a free interchange of birds from one area to the other. The Pottawattomie and Ottawa areas consist of 2,800 acres, principally marshy bottomlands with adjacent timbered areas. These areas serve as both private and pub- lic hunting grounds. The Swan Creek Wildlife Experiment Station has a 550-acre sanctuary of par- tially flooded land, once farm land, and timbered bottomland. The Todd Farm Sanctuary comprises 1,500 acres of drained lake-bottom farm land. This sanctuary furnishes both feed- ing and resting sites. Hutchins Lake, north of the farms, is used by geese as a rest lake in the autumn. A spring-fed creek crossing the farm remains open through the winter, and food is available to the geese in the cultivated fields. The greatest concentration of geese re- corded in the above sections was 6,000, in the autumn of 1945. The wintering flock was estimated at 2,000. Both figures are said to represent spectacular increases in comparison with those of previous years. In 1944 the wintering flock was estimated to be only 400. 3. A 250-acre sanctuary at Leidy Lake in Leonidas Township, St. Joseph County, fig. 39Fj serves as an important spring concentration point ; over 2,000 geese were estimated to be on the area in 1946. It is used less in the autumn, 300 to 400 being average numbers of geese present at that time. Saginaw Bay.—Saginaw Bay is a major concentration area for both autumn and spring flights of Canada geese, fig. 39G^. The spring flights may consist largely of South Atlantic geese en route north from the Miner Sanctuary. The geese do not linger long at Saginaw Bay in the autumn because of the absence there of sanctuaries. They have been forced by hunting pressure to reverse their normal March, 1950 Hanson & Smith: Canada Geese 121 daily routine, feeding after dusk in the grain fields and roosting in daylight hours on the open waters of the bay. They do not winter in this sector, as there is no open water. The best estimates axailable place the maximum numbers frequenting the bay during the autumn or spring migrations at about 15,000 birds. Leelanau and Benzie Counties.— Leelanau and Benzie counties in Michi- gan, fig. 39//, constitute a less important concentration area than the Kalamazoo River or Saginaw Bay areas. A number of scattered sites are favored : Glen Lake and about four sections of hilly grassland in Empire Township, Leelanau Countv, and Lake Ann, Upper Platte River, and Platte Lake in Benzie County. A 1,200-acre ref- uge recently established in Empire Town- ship provides feeding and resting areas, Canada geese have used this area as a regular stopping place for only about 10 years. In recent years as many as 2,000 geese have frequented it regularly, but few winter in the area; there were about 50 in 1945. Some Canada geese may nest in this region. Other Michigan Areas.—There are two other concentration areas in Michigan of less importance than the above. The Seney National Migratory Waterfowl Refuge, fig. 39A', consists of 30,000 acres surrounded by a vast area of wild land. The autumn concentration in 1945 was estimated at 3,000 geese. The spring maximum was 2 500. The Alpena Sanctuary, fig. 39L, com- prises 500 acres of land on the Thunder Bay River in Alpena Township, Alpena County. Geese stocked at this refuge have attracted as many as 400 migrants in the autumn. Wisconsin In Wisconsin there are two refuges or concentration areas of importance. Rock Prairie Refuge.—The Rock Prairie Refuge, fig. 39jM , consisted of 640 acres when established in 1936. Be- fore the refuge was relocated, it lay part- ly in Richmond Township, Walworth County, and partly in Johnstown Town- ship, Rock Count). In 1945 the refuge was shifted 3l j miles to the west so that it lay entirely within Rock County. The entire refuge is in cultivated prairie uplands and is used for feeding only. Since feeding was initiated in 1940, between 25 and 45 tons of corn have been fed each season. The geese that frequent this refuge in the autumn and winter usually fly to Lake Geneva and Lake Koshkinong for roosting. Canada geese do not remain in southern ^Visconsin during severe winters. In 1945—1-6, local estimates placed the win- tering flock at 3,000. In 1942, 4,500 geese wintered in these two counties (Zimmerman 1942). Peak autumn popu- lations in Rock and AValworth counties have generally varied from 4,000 to 6,000 birds. The btnld-up in numbers of geese in the autumn of 1942 is shown graphically in fig. 38. Greenwood Farm Refuge.—The Greenwood Farm Refuge, established in 1940, contains 1,751 acres. It is situated in Hancock and Deerfield townships, western Waushara County, fig. 39A^ Altiiough it is intended primarily as a rest area, in some parts of this refuge farmers are paid to leave corn standing in the fields for the geese. The flocks roost on the sand bars of the Wisconsin River, about 20 miles to the west. The refuge was first used by a few geese in 1942; as many as 3,000 birds had been reported on the area by 1946. Ohio Lake St. Marys, or Grand Reservoir, fig. 39/*, a 17,500-acre impoundment in Mercer and Auglaize counties, is heavily used by Canada geese in migration, but few geese winter there or in other parts of Ohio. January inventory figures, 1941—1946, show an average of only 400 Canada geese in the entire state; inventory figures for the state, 1936—1941, averaged 1,600 per year. Indiana Hovey Lake, fig. 390, the most im- portant wintering area for Canada geese in Indiana, is located in Posey County in the extreme southwestern tip of the state, 4 miles from the confluence of the Wabash and Ohio rivers. The lake and adjoining marsh and swamp land, totaling 900 acres, were purchased in 1938 with funds made available through the Federal Aid 122 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 in Wildlife Restoration Act. The lake, nearly a half mile wide and three-quarters of a mile long, has an area of about 400 acres. Bald cypress is found around cer- tain parts of the lake, but willow, elm, and soft maple, with an understory of buttonbush, occupy most of the shore line. Approximately half of the lake is open to waterfowl hunting; the remainder is a refuge. No supplemental feeding is carried out on the refuge sector, a factor that may partially explain why the geese that use the area have retained their wild- ness. Nearby wheat fields are cropped by the geese to a considerable extent, but no serious damage has been reported. During the winter period the geese do much of their feeding in overflow, bottom- land cornfields that have been harvested with mechanical pickers. There are seldom more than 300 geese in the vicinity of Hovey Lake during the hunting season. Maximum numbers during three recent winters are as follows: 1,000 on January 8, 1944; 1,500 on January 20, 1945; and 2,000 on Januarv 27, 1946. Arkansas The geese wintering in the lower White River and Arkansas prairie area, fig. 395, use several distinct types of habitat : the flood-plain swamp lakes of the White River National Wildlife Refuge, the neighboring prairie area of Arkansas County, and the sand bars of the lower Arkansas River and adjacent parts of the Mississippi River. The flood-plain lakes are shallow, cypress-rimmed oxbows, devoid of sub- merged vegetation and used by the geese only for roosting. The geese make daily flights from these lakes to the prairie for feeding. The prairie is intensively culti- vated ; rice, winter oats, soybeans, and lespedeza are the principal crops. The practice of leaving the rice fields fallow periodically and using them for pasture makes attractive foraging areas for geese, as the ground between the old rice levees is frequently flooded or at least wet during the winter. Because of the difficulty of censusing the extensive areas of bottomland swamps, we believe that in most years our data on populations in Arkansas are not reliable. In the winter of 1943—44, the population was successfully censused and estimated to be 5,000, but we are unable to state with any certainty what the population was in prior or subsequent years, as the birds win- tering in this region occasionally use the Mississippi River bars and may have been included in the estimate for the Mis- sissippi River area. Duplications in the inventory figures for Arkansas and the lower Mississippi River in 1944—45 may account for the indicated increases for this region in that winter and partially ex- plain the apparent sudden great drop in the total population of Mississippi flyway geese in the following year, table 7. Lower Mississippi River Islands and bars in the Mississippi River attractive to wintering Canada LAKE CHARLES 5 GRAND LAKE 6 CALCASIEU LAKE 7 SABINE LAKE 8 GALVESTON BAY 9 AVERY ISLAND 10 LAKE FAUSSE POINTE ^'^ 4S.—Location of wintering grounds of Canada geese on the coast of the Gulf of Mexico. March, 1950 Hanson & Smith: Canada Geese 123 geese extend north to Chester, Illinois, and south to White Castle, Louisiana. However, except for scattered flocks, few Canada geese have wintered on the Alis- sissippi River between Cairo, Illinois, and the Tennessee-Mississippi state line in recent years. The portion of the Mississippi River used by Canada geese throughout the win- ter, fig. 39R, has an area, from levee to levee, of well over 1,500 square miles. In this huge expanse of territory the channel has constantly shifted by cutting and tear- ing on one side and depositing on the other ; the result is a labyrinth of chutes and oxbows that have formed numerous islands and bars. Each island usually has one or more sand bars, and most of the bends in the channel have bars on the inside, fig. 44. Portions of the higher is- lands and bars, covered with small switch willows, grasses, and sedges, are used as feeding areas by the geese. At times, the bark of the small switch willow appears to be staple food of geese throughout the area. The geese using this section of the river are widely scattered ; usually they are in small or medium-sized flocks, but occasionally in large flocks. They show a preference for certain bars, which they use year after year. Varying water stages affect the accessibility of the bars to the geese and may cause the flocks to shift about when water levels change rapidly. In primitive times the geese depended on forage produced on sand bars and in shallow flood-plain lakes, since there were then no cultivated crops in the bottom- land country. Early agricultural devel- opments tended to keep them on the bars in the southern sections of the flood-plain and upper delta country because cotton and sugar cane were the only crops ex- tensively raised. Each year increasingly large acreages are planted to winter grains and legumes in the north and central por- tions of the bottomlands, thereby increas- ing the food resources for the geese in that section. The habitat at Grand Lake and Lake Fausse Pointe, fig. 45, while near the coast, is of the sand-bar type rather than marsh. The geese are found on the upper ends of the lakes where the distributaries of the Atchafalaya River have forjned a subdelta, creating conditions very similar to those found on the river sand bars. The geese sometimes work back and forth across the Atchafalaya swamp between Grand Lake and the Mississippi River, as less than 25 miles separate the lake from White Castle, Louisiana, the nearest point on the river. Our data on goose populations in this sector of the valley are meager. Esti- mates made by the United States Fish and Wildlife Service have varied from 1,600 to 10,000 geese between 1944 and 1946. Accurate census figures are especially 124 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 ncedi-d from the lower Mississippi River each year, hecause data from this area are apt to indicate to what extent kills made in the Horseshoe Lake region are at the expense of populations wintering below Cairo, Illinois. Coastal Marshes The coastal marshes of Louisiana and east Texas extend from the mouth of the Mississippi River west to Galveston Bay. In this vast expanse of marsh, totaling over 5,000,000 acres, less than 700,000 acres are inhabited by Canada geese. Western sections of this range are used also by white-fronted geese, and between the Delta and Rockefeller refuges the winter range of the Canada goose is over- lapped by that of the blue and snow geese. In the coastal marshes are three prin- cipal concentration areas for Canada geese, figs. 39 and 45, and, as these vary somewhat as to type of habitat involved, each is discussed separately. Delta of the Mississippi River.— At the mouth of the Mississippi River, Canada geese are concentrated on the Delta National Wildlife Refuge and the adjacent area in the vicinity of Main Pass, T in figs. 39 and 45. Here they use a variety of marsh types, from the rel- atively hard deltaic flats bordering the Gulf Coast to the deep marsh—the "floating prairie" of the interior. This is the most isolated wintering area on the Louisiana section of the Gulf Coast. The marshes to the west between the Delta Refuge and White Lake, an air-line distance of 180 miles, are devoid of Canada geese, except for a small flock inhabiting Avery Island. White Lake and Lacassine Ref- uge.—In the White Lake and Lacassine area, Canada geese occupy an extensive range (U in figs. 39 and 45): east to Cow Island, north to the edge of the prairie below Gueydan, west to Sweet Lake, and south to the Rockefeller Ref- uge, which lies below Pecan Island; the range does not include Grand Lake and Lake Misere. Within this area Canada geese are most abundant south of Guey- dan, where prairie and marsh merge, and on the Lacassine Refuge. In the zone of contact between the prairie and the marsh, there are marginal rice fields and wet pastures, interspersed with patches of maidencane, Panicuni heiiiitOTnon, and southern wildrice, Zizaniopis miliacea. A vast expanse of maidencane and Jamaica saw grass, Mariscus jaimaicensis, with occasional low grassy ridges, is found throughout the marsh between White Lake and the edge of the prairie. At the west end of Pecan Island there are old stranded beach ridges roughly paralleling the coast line. The ridges, pastures, and rice fields are used extensively by geese for feeding areas. The deep marsh is used primarily for roosting. Calcasieu Lake to Galveston Bay. —The Canada geese occupying the range from Calcasieu Lake to Galveston Bay are found on Calcasieu and Sabine lake ridges, the edge of the prairie, and the relatively high sea-rim marshes from Johnson Bayou to Port Bolivar, Texas, {Y in figs. 39 and 45). The Louisiana section of this range is limited, consisting only of a narrow fringe around an exten- sive area of deep marsh. In Texas, how- ever, the reverse is true: a wide area of sea-rim and prairie marshes around a rel- atively small area of deep marsh. Con- sequently, almost the entire Texas area is good Canada goose range. The marshes in this area, along with the high marshes of southwestern Louisiana, are heavily grazed by cattle, which keep the forage in an ideal condition for feeding geese. The geese frequently roost on the inshore waters of Calcasieu and Sabine lakes and Galveston Bay, as well as on such smaller water areas as Black and Brown lakes. Inventory of goose habitat on the Gulf Coast in 1943-44 and 1944-45 revealed populations of 12,000 and 13,000 birds, respectively. Partial coverage of the Gulf Coast in 1945—46 indicated no significant change in numbers over the previous 2 years. The above figures represent great divergence from what was commonly be- lieved to be the Canada goose population on the Gulf range. Vast areas of ex- cellent marsh are unused by Canada geese. Alfred M. Bailey stated (personal com- munication) that, even in the late twenties, Canada geese could be found in only a few places on the Gulf Coast. The geese frequent these same places today. "It has become scarcer of late years," Bailey & Wright (1931) wrote several March, 1950 Hanson- & Smith: Canada Geese 125 years ago regarding the Canada goose population on the Gulf Coast. While there may have been much greater num- bers of Canada geese wintering in the marshes of Louisiana 25 or more years ago, the decrease to present-day popula- tions has not occurred altogether in recent years. It is more likely that the decrease was a gradual one, probably much of it caused by heavy shooting in northern parts of the range and in Louisiana. It seems altogether probable that at least some of the geese that normally would have wintered in Louisiana have been de- coyed into Horseshoe Lake for entire seasons and have contributed to the annual kills there, but data are not available to show the e.xtent to which hunting at Horseshoe Lake has affected Gulf Coast populations. For reasons discussed under "Autumn Migration Routes," it is diffi- cult to believe that the kill made at Horse- shoe Lake in any recent year would materially affect Louisiana populations the same year, for there are no data to show that an appreciable turnover in the population occurs at Horseshoe Lake within a single season. In 1943—14, aerial coverage showed the following distribution of the Gulf Coast Canada goose populations: Delta Refuge I OOn, White Lake and Lacassine Refuge 7,000, Sabine Refuge (Gum Cove and Hackberry Island) 4,000. GOOSE BEHAVIOR AND HUNTING LOSSES The tremendous number of Canada geese bagged in the vicinity of Horseshoe Lake in recent years has made this area one of the most widely publicized shoot- ing spots on the continent. The fearless and unwary behavior of the geese that winter at the Horseshoe Lake Game Ref- uge is responsible in large measure for the heavy kill, fig. 46. The response of this flock to hunting is contrary to the traditional reputed behavior of Canada geese. For centuries, the Canada goose has been e.xtolled as one of the wisest and wariest of all birds and has been regarded as one of the most difficult to hunt success- fully, but hunters and personnel engaged in wildlife management who have ob- serxed the habits of the Horseshoe Lake flock in Alexander County agree that these habits do not conform to the tradi- tional pattern of Canada goose behavior. How can the behavior of the Canada goose in Alexander County be reconciled with its traditional reputation ? If the species is so wary or intelligent, why is it so unsuspicious and easilv killed in Alexan- Fig. 46.—A portion of the Horseshoe Lake Canada goose flock near the refuge headquarters. In many years, when food was scarce this flock lost much of its normal wariness. 126 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 der County? That the traits of the Horseshoe Lake flock are apparently singu- lar cannot be denied, but there are many clues in the literature that help to explain its seemingly perplexing behavior. For example, many authors, after discussing tile sagacity of the Canada goose, cite examples of the behavior of this goose that conflict with their previous remarks. Grinnell (1901) has aptly expressed the enigmatic behavior of Canada geese: "The wild goose has long been proverbial for his shyness and wariness, and he well deserves the reputation that he has gained, and yet sometimes he is found to be 'as silly as a goose.' So that the gunner who follows the geese enough to see much of them will find that at one time great acuteness and at another a singular lack of suspicion are present in the ordinary wild goose. Few birds are more difficult to approach than these, and yet few come more readily to decoys or are more easily lured from their course by an imitation of their cry." A veteran goose hunter de- scribes the Canada goose as "a bird of man} moods. At times, very wise, but at other times very foolish" (Darby 1916). Barnston (1862), referring to the Canada goose in the Hudson Bay region, writes: "Its disposition has less of wild- ness in it than that of the snow goose." These citations and others given below show that many of the traits which make Canada geese vulnerable to hunting have been recognized elsewhere in the country, indicating that the behavior of the Horse- shoe Lake flock is not as unique as one might be led to suspect. The unusual aspect of the reactions of the Horseshoe Lake geese seems to be that all or most of their behavior traits that tend to make them vulnerable to hunting are exhibited in the vicinity of Horseshoe Lake. Wariness, Innate and Acquired Many observers point out that geese are not so wary as various species of ducks, especially the mallard and black duck. Brandt (1943) noted a difference in wariness even in the newly hatched : "Young ducks of most kinds, just hatched, are very wild little creatures, which scatter at once and hide by all sorts of ruses. Newly hatched geese are most trusting little fellows," The origins of the behavior differences between ducks and geese are deeplv rooted. Lorenz (1937), Lack (1941), and Tin- bergen (1942, 1948) have contributed to an understanding of these origins, which seem to relate in an important degree to the "innate perceptory patterns." There appears to be an inverse relationship be- tween the specificity and specialization of these patterns and the degree to which the behavior patterns are ( 1 ) directed by "imprinting" (Lorenz 1937) during a brief period after hatching and are (2) developed, subsequent to the imprinting stage, by associative learning. The acute wariness that adult geese normally possess seems to be mostly an acquired trait. Experience and associa- tion of the young geese with older birds appear to play an important role in the development of the traditional behavior pattern. If newly hatched goslings are taken before they have left the nest and are hand reared, their subsequent be- havior shows considerable divergence from that of the wild birds. The readiness with which the Canada goose will become semidoniesticated when given protection may possibly be related to the slow devel- opment of wariness in young birds. A factor contributing to the fearless behavior of the Horseshoe Lake geese is the dual role played by man on and in the vicinity of the refuge. As the geese are accustomed to the sight of refuge workers, visitors, and the activities of a relatively dense rural population outside the refuge from the time they arrive in the autumn until the opening of the hunting season, they are apparently unable to comprehend the unfriendly role of the hunter. The same reaction to man has been found to be true in other places. Todd (1940) writes, "Under the protection now afforded at Erie Bay, the geese are less wary; on March 25, 1932, a party of which I was a member saw about twenty- five resting on the shore of a sheltered cove, and without apparent concern they per- mitted us to drive up in an automobile within one hundred feet." Stone (1937) relates how Canada geese have responded to food and pro- tection on the Atlantic Coast: "In season the farmers of this region [Cecilton, Maryland] go goose shooting on the March, 1950 Hanson & Smith: Canada Geese 127 wheatlields and have decoy Canada geese to attract the wild birds. Of late years the ground has been baited and the geese return year after year to the places where they have been fed, which accounts for their abundance and tameness. " (Notes from a field trip taken in February, 1927, when decoys and baiting were permitted.) Even when Canada geese are winter- ing along the vast coastal marshes of the Gulf of Mexico, where, with an abun- dance of natural food, they might be ex- pected to retain their independence, free- H\ing wild individuals will momentarily accept man at close range. At the Florence Club, near Gueydan, Louisiana, geese formerly used for decoys and tame cripples are brought into the club grounds for feeding each evening by calling and beating on a tin pan. On these occasions, numbers of wild geese accompany the tame birds into the club grounds and feed from the caretaker's hands. At all other times these same individuals seem to be un- approachable. The importance of the role of man in conditioning the behavior of an entire flock was brought forcibly to our atten- tion at the Miner Sanctuary. Until about 1925, wild Canada geese using the sanctuary were fed at a pond, 150 feet in diameter, which is located a few yards from the secondary road that passes in front of the Miner home. During the migration periods, wlien the geese were fed at this small pond, they were usually under the observation of large numbers of visitors, who, unconcealed, viewed them at close range. As a result of this en- couraged familiarity, the vigilance of the geese toward man relaxed to such an extent that the local kills increased. Be- cause the situation needed to be remedied, the geese were fed at a larger pond, fig. 40, away from the road, where they were hidden from public view by a dense grove of pine trees and a tight, 7-foot, wooden fence. V^isitors who wanted to view the main concentration were required to use blinds or an observation tower overlooking the ponds. The resultant change in the behavior of the geese was profound, and local kills were soon reduced. After these new management measures were insti- tuted, the sight of man was usually sufficient to flush the geese ; previously, they had to be practically driven out of the front pond before they would take flight. Although the Canada goose possesses mental powers that at times seem to be superior to those of most birds, and that are undoubtedly of great survival value under primitive conditions, individuals appear unable to solve problems of self- preservation that arise in a highly modified environment such as that in the Horse- shoe Lake region. During the hunting season the geese wintering in that region exhibit almost a complete disregard for gunfire, flying back day after day to fields that often are the most heavily shot. 'Fhis situation has perhaps been aggravated in recent years b\ the fact that the geese can feed in these same fields with impunity after the close of the day's shooting but are shot at on returning to feed the next day. The flock as a whole appears to be baffled by the presence of food and pro- tection on the refuge at all times, and by the presence of food (standing corn, win- ter wheat) at all times but protection only a part of the time away from the refuge. Family Grouping Jenkins ( 1944) , in a report on the social organization of a family of geese, states that "This well-integrated [Canada goose] family might be called a family supra- organism, since it performs the activities of a larger, more complex individual, through coordination of its components. This results in the dominance of the family, which is of survival value to its members in that they can feed first and rest in the center of the aggregation and are not pecked or chased." Strong family ties in geese are un- doubtedly of survival value against natural enemies, each family being a protective unit. Against man, during the hunting season, family grouping proves to be a liability, as the death or injury of one member frequently lures the rest of the family within gun range. Many a veteran goose hunter can cite examples of sur- viving members of a family flock, con- fused by the loss of one of its members, returning to a shooting pit to be shot at again. Bent (1925), in describing the duck-stand method of shooting geese on the inland ponds and lakes of Massa- 128 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 cliusetts, writes, "When the geese are near enough and properly bunched a raking volley from a battery of guns is poured into them and other shots are fired as the survivors rise, with the result that very few are left to fly away. Even some of these may return and be shot at again if the leaders or parents of the young birds have been killed." This behavior trait has also been reported by Phillips (1916) : "Now if a successful shot [probably mean- ing a series of shots fired at one time] is finally made into such a flock, and per- haps one half or three fourths of their number have been killed, the remainder, after a few turns in the air, or a short flight of five or ten minutes, will almost always return to the pond, where, if not actually disturbed, they will remain from several hours to a day or so. Sometimes they will decoy a second time." The closely allied little cackling goose, Brarita riiiniina, sometimes exhibits the same type of behavior. "Even upon first arrival [in Alaska] many of the birds appear to be mated, as I have frequently shot one from a flock and seen a single bird leave its companions at once and come circling about, uttering loud call- notes. If the fallen bird is only wounded its mate will almost invariably join it, and frequently allow itself to be ap- proached and shot without attempting to escape. In some instances I have known a bird thus bereaved of its partner to re- main in the vicinity for two to three days, calling and circling about" (Nelson 1887). Because of the concentration of birds at Horseshoe Lake, individuals from a broken family that return to the shooting fields in search of missing members can seldom be identified. However, one such instance was observed by a hunting club owner in 1944. Four geese from the ref- uge swung over a pit and two were dropped ; the two remaining flew back toward the refuge, and when over the lake they made a wide swing and came direct- ly back over the same pit, where they also were shot. In 1945, another incident was noted that demonstrated the high vulnerability of the surviving members of a broken family. A flock of five geese entered a shooting field and, as the birds approached the second pit, two of the flock were killed and one crippled. The two uninjured geese immediately alighted and remained with the cripple for about 10 minutes be- fore taking flight toward the refuge lake. On their way to the lake, they were crippled, one being hit so severely it bare- ly gained the refuge. It is apparent from these examples of Canada goose behavior that the perma- nency of family ties offers one explanation why geese, unlike ducks, cannot easily be shot out of a field. Surviving members of broken families searching for mates, parents, or young that have been shot probably contribute appreciably to the total bag; thus, a high kill at a shooting club early in a hunting season may in- sure continuance of a high kill through the remainder of the season. The pres- ence of the survivors over the shooting fields would tend to decoy unbroken families into gun range. As a result, the performance of the geese at some clubs toward the close of the hunting season might be aptly described as a perpetual- motion shooting gallery, the birds moving across the hunters' horizon in a never- ending procession against the heaviest kind of gunfire. Sociability The Canada goose is a social bird and, except during the breeding season, it tends to congregate in fairly large numbers. This tendency, which was common to some of our now extinct species of birds and mammals, often has two im- portant undesirable results: first, under some conditions it causes the species to lose some of its normal wariness ; and, second, when the remnants of a popula- tion band together they give an un- warranted impression of general abun- dance. Audubon (1843) made the observation that the behavior of geese using small water areas may differ from that of flocks that resort to large bodies of water ; that is, the behavior may vary according to relative densities on an area. "The Canada goose is less shy when met with far inland, than when on the sea-coast, and the smaller the ponds or lakes to which they resort, the more easy it is to approach them." March, 1950 Hanson & Smith : Canada Geese 129 Apparently wariness is related both to the total size of an aggregation and its size in proportion to the area it uses. The first relationship may be of a psychological nature ; many species of mammals and birds show a reduction in wariness when they are in large herds or flocks. It is fairly common knowledge that many species, for example the ruffed grouse, are very wild when at the bottom of their cycles but are quite readily killed when abundant. At Horseshoe Lake the wari- ness of the geese in the autumn decreases a:> the flock increases and spreads out over the refuge, thereby reducing the area of unoccupied ground to which disturbed flocks can retire. \Vhile the loss of natural wariness in aggregations of wild game is serious from a long-term standpoint, the impression of abundance that local concentrations create in the minds of observers may serve as a fairly immediate threat to the future of a species since it becomes a premise for un- limited gun pressure. To substantiate this point we need only cite recent his- torv of the flock at Horseshoe Lake. From 1942-43 to 1945-46 this flock had grown smaller each year, while most of the local residents and visiting hunters at Horseshoe Lake believed that each year there were "more than ever." To many hunters, a closed season on this flock in 1946 seemed to be a needless infringement of their privileges. Hewitt (1921) has stated, "It should also be pointed out that when a formerly abundant animal becomes reduced in numbers the remnant may tend to herd together and thus give an impression locally of great abundance Local abundance, therefore, should never be taken as an indication of general abun- dance, and as a reason for permitting kill- ing in large numbers." Jackson ( 1943) has stressed the dangers of overshooting local remnants: "Extinc- tion in ever)- case was probably brought about at first by gradual depletion of the population and through local extirpation. When the population becomes reduced to a danger point, extinction may come with unexpected rapidit\-. Dislike the asser- tion as we may, in recent times the human species has been the prime factor in the extermination of other species." HISTORY OF GOOSE HUNTING IN ILLINOIS The hunting of Canada geese was once common in widely scattered areas over the state of Illinois. In most of the areas that formerly offered considerable shoot- ing, the hunting of Canada geese as a sport of any consequence has ceased to exist. In a few, goose hunting has con- tinued on a smaller scale; only ir. the Horseshoe Lake area has the kill in most years been high. Because the histor\- of the sport in Illinois parallels the historv of many other goose-shooting areas in the flywa\-, and because it relates to present goose-management problems, it is briefly reviewed here. Two factors have been chiefly respon- sible for changes in the methods of goose hunting, and for the decrease or increase of goose hunting in different sectors : ( 1 ) the development of state, federal, and pri- vate refuges, frequently attended by arti- ficial feeding, and (2) the outlawing of both baiting and use of live decoys. For- merly, fair bags of Canada geese were made on the Big Foot Prairie in the north- eastern portion of Illinois near the Wis- consin state line, but, with the establish- ment of a refuge and feeding station in southern Wisconsin, fewer birds have been a\ailable to northern Illinois hunters. The Putnam area, west of Lake Senach- wine, in the Illinois River valley, yielded fair bags of geese until 1935, when both baiting and use of live decoys were pro- hibited. When feeding was curtailed, the area no longer proved attractive enough to hold flocks for sufficient time to pro- vide hunting. In about 1925, i\Iason County, border- ing the Illinois River, was the most im- portant goose-shooting area in Illinois. 1 he use of live decoys in the fields of winter wheat situated near large bottom- land lakes was responsible for the popu- larity of this area. Field-pen hunting of Canada geese at private shooting clubs and at commercial day-shooting "clubs" in this county was centered largely east of Clear and Chautauqua lakes, northeast of Havana, and between Bath and Sni- carte. The average kill of honkers in the Clear Lake area in the twenties is reported to 130 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Fig. 47.—Before pavcti roads and the Horseshoe Lake CJame Refuge brought the hunter auti the Canada goose into close proximity, goose hunting in southern Illinois was a fairly arduous undertaking. Here is a party of well-equipped hunters on their way to a Mississippi River bar. This photograph was taken in Alexander County in the early twenties. (Photograph by Bob Becker.) Fig. 48.—Canada goose hunting as it was carried out on the Mississippi River sand bars in southern Illinois before the creation of a refuge at Horseshoe Lake. (Photograph by Bob Becker.) March, 1950 Hansox & Smith: Canada Geese 1,51 have been about 100 per year or roughly 7 to 10 per cent of the number reported to have lingered in that area in those years. The top kill in the Bath-Snicarte aiea by the Brownstone Club in 1928 was 514 geese, more than the combined kill of all the other clubs in that region. In the late twenties and early thirties the average kill at Brownstone was about 400 per year. After the prohibition of baiting and use of live decovs, commercial dav- change in the type of wintering habitat, from one that was relatively primitive to one approaching parklike conditions, goose- hunting methods underwent an equally drastic change. The following description is quoted from an unpublished report in 1941 by Arthur S. Hawkins, then of the Illinois Natural History Survey, and Paul S. Smith, then, as now, of the United States Fish and Wildlife Service. Fig. 49.—Scene at goose-hunting club near Horseshoe Lake. The Horseshoe Lake region has been one of the most intensively hunted areas in the United States. The refuge totals only about 3,700 acres, but between 1941 and 19+5 the area around it devoted to hunting averaged 11,000 acres controlled by an annual average of about 50 clubs. The number of pits and blinds in this acreage in the same period averaged appro.\imately 400 and the total hunter capacity of the area 1,000. shooting in the Illinois River valley was at an end, and only one private club primarily for goose shooting still exists. The continuance of Canada goose hunting in Mason County is due largely to the operation of two refuge areas, one private and one federal, that holds the birds in the area. In recent years, kills in the en- tire Illinois River valley have been about 400 birds per hunting season. The river bars and islands of the Mis- sissippi River between Chester and Cairo, Illinois, have been a wintering ground for C anada geese for many years, and since pioneer days this area has been noted for the goose shooting it afforded. The recent concentration of geese at Horseshoe Lake is in marked contrast to the wide dispersal of the birds in earlier tiines. With the At the beginning of the present century there were comparatively few goose hunters, because goose hunting was no sport for the novice. Most of the hunters were skilled river men ; those who traveled to the hunt- ing; grounds by land did so by horse- or mule-drawn vehicles over many tiresome miles of nearly impassable roads, fig. 47. Once at the shooting grounds there remained the task of digging a pit and placing the decoys, fig. 48. After a hard day's hunt, the hunter either camped out on a bare sand bar or faced a long return trip. Al- though there were more geese and fewer hunters in those early days, real skill was required to bag geese consistently because the goose range was extensive and the sand bars numerous. Then, as now, silhouettes or "shadows," as they are called locally, were used to de- coy the geese. Live decoys were seldom 132 Illinois Natural History Survey Bulletin Vol. 25, Art. ."? Fig. 50.—Modern-day goose hunters in a typical pit at a day-shooting club adjacent to the Horseshoe Lake Game Refuge. Fig. 51.—Typical goose blind in a soybean field near the Horseshoe Lake Game Refuge. March, 1950 Hansox & Smith: Canada Geese 133 used until after 1906, when it became the custom to use three live decoys in combina- tion with the silhouettes. The silhouettes were arranged in V-formation, with the apex of the V downwind from the pit. A live "caller" was placed at the vertex and at each end of the V. In between were the "shadows." Bait was not used, but, in order to induce the geese to work into the proper bar, hunters sometimes placed "scarecrows " on adjacent bars. Improved roads and faster transportation brought goose hunting within the reach of the masses, fig. 49. Heavy competition for the better hunting places ensued. The de- mand for more hunting grounds resulted in the development of field shooting. Long before baiting came into promi- nence, goose hunters recognized that no other type of feed was more attractive to geese than a large field of fall-planted wheat or rye. As soon as the weather turned cold, however, shelled and ear corn, wheat ker- nels, cowpeas, and similar feeds, when properly scattered, proved very attractive to the geese, although their desire for greens continued. When decoys were used, the usual proce- dure was to construct a pen using a roll or two of 3-foot wire. In this pen were placed as high as 100 geese. L^sually one or more geese were separated from their mates so that they would "talk" back and forth to each other. Another trick was to place a trained goose, which was wing-clipped, in the blind; the goose was then thrown from the blind and permitted to walk to the pen, "talking" to its mate in the pen as it went. If the first decoy failed to entice a wild flock within the range of the gunner, others were released from the pit until the wild geese decoyed as desired, or the supply of decoys was exhausted. Only a small per- centage of captive geese behaved in such a manner as to make good decoys. These geese became as valuable an aid in goose hunting as well-trained bird dogs are in quail hunting, and commanded equally high prices on the market. The function of live decoys was to attract the geese, while that of feed was to hold them and to encourage the birds to return again. One answer to increased hunting pres- sure was the formation of goose hunting Table 10.—Goose hunting regulations as they applied to Alexander County, 1927-1945. Illinois, Year 134 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 clubs, but since the time of the Egyptian Hunting and Fishing Club, organized in 1904, goose clubs have changed considerably in Alexander County. Present-day clubs, with a few exceptions, are strictly commer- cial. In contrast, this first club (which had annual dues of only $5) was a nonprofit organization. At one time, it boasted a membership of 50, all local hunters. In 1941, there were at least two dozen clubs in Alexander County, each of which, ac- cording to a direct comparison of kill rec- ords, killed more geese annually than did the Egyptian Club. Goose hunting first took on a commercial aspect when in 1913 a Chicago business man began to lease the sand bars most fre- quently used by the geese. By 1916, most of these bars were no longer open to public hunting. Up to that time field shooting had been scorned by most real goose hunters. Now that the river shooting was largely under the control of a few men it was field shooting or nothing for the old timers. The purchase of Horseshoe Lake for a refuge in 1927 created a boom in commer- cialization of goose shooting. Mediocre farm lands located near the refuge suddenly commanded fancy prices. Now almost every field located around the refuge contains pits and blinds during the hunting season, figs. 50 and 51. Data obtained from veteran hunters on the number of geese killed along the Mis- sissippi River in the eighties and later have been too contradictory to permit any de- finite conclusions. None of the informa- tion obtained, however, indicates that the kills made in those early years exceeded recent kills at Horseshoe Lake. A sum- mary of hunting regulations as they applied to geese in Alexander County is given in table 10. The relationship between number of hours of open season, number of geese bagged, and hourly kill per season is shown in figs. 52 and 53. Neither the hourly bag nor the seasonal bag shows significant correlation with the number of hours open to hunting. o 180 - HOURLY BAG OF GEESE - HOURS OPEN TO HUNTING •o o a 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 YEAR Fig. 52.—Hourly bag of Canada geese and number of hours open to hunting in Alexander County, Illinois, 1927-1945. March, 1950 Hansox & Smith: Canada Geese 135 S 10000 i, 5000- CD i 4000- 3000 2000 1000 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 YEAR Fig. 53.—Seasonal bag of Canada geese and number of hours open to hunting in Alexa County, Illinois, 1927-1945. ANNUAL BAG In recent years the annual kill by hun- ters of Mississippi flyway Canada geese has probably exceeded losses resulting from any other single cause. The hunter kill includes the geese bagged and those so severely crippled by gunfire that they soon die. Hochbaum ( 1'544) has pointed out that the ratio of the number of hunters to ducks is such that it is mathematically possible for the licensed hunters legally to exterminate the continental duck popu- lation in one season. In the case of the Horseshoe Lake flock of Canada geese, the threat of extirpation has been real. If hunting in Alexander County, Illinois, had been permitted for the duration of the full 80-day waterfoul season either in 1944 or 1945. that population might have been reduced to a remnant, fig. 54. Of all mortality factors, the bag by hunters is the one that can be most easilv controlled to insure preservation of the Canada goose population. Insofar as management of Mississippi flyway Canada geese is concerned, the annual bag has two subdivisions: the bag made by. In- dians on the breeding grounds in Canada, and the bag made by hunters in southern Canada and the United States while the geese are in migration or in the vicinity of the wintering areas. On Breeding Grounds Man is believed to be the predator taking the heaviest toll of Canada geese on the breeding grounds. Responsible for the bulk of the take in the James Bay area are the Cree Indians, natives of the region ; the handful of white residents also kill a few geese. Food is the primary con- sideration for killing geese in the North ; any sport involved only adds flavor to the undertaking. In the James Bay and Hudson Bay area the native populations are dependent on 136 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 waterfowl as one of the few- reliable sources of meat; big game animals are usually scarce and small game is subject to violent cyclic fluctuations in numbers. '1 he importance of waterfowl, particular- ly geese, to the Cree Indians in former no data regarding Churchill, but we con- clude from Barnston's report that the white-fronted goose was shot in fair num- bers by the Churchill Indians. It seems certain that snow geese and lesser Canada g;ese also contributed to the total kill in iit .r^mcrivt*y Fig. 54. — The registered kill of Canada geese (1,400) on opening day at hunting clubs near Horseshoe Lake in 1945 was approximately equivalent to the number of geese shown in this illustration. (Photograph taken at the Horseshoe Lake Game Refuge by George W. Sommers.) years can be readily realized from Barns- ton's report (1862). He estimated the annual kills of all species of geese on the west coast of James Bay and the south coast of Hudson Bay as follows: Moose Factory district, 10,000 annually; York Factory and Churchill district and region to the north, 10,000; Fort Albany district, 17,000 to 20,000 in the autumn and 10,000 in the spring; Fort Severn district, 10,000. The species of geese that made up the bag at these posts must have varied con- siderably then as they do today. We have that area. In recent years at Moose Factory, Fort Albany, Attawapiskat, figs. 55 and 56, and VVeenusk, the annual kill of geese has consisted chiefly of blue geese and snow geese, with Canadas running a poor third. At Fort York, the annual kill of Richardson's geese, Branta hut- chmsii, equals the combined kill of snow, blue, and Canada geese ; the Canadas are outnumbered in the native hunter's bag at this post by the "wavies." Big game represents an unpredictable source of food for the present-day Indian. In the early part of this century, caribou, March. 1950 Hanson & Smith: Canada Geese 137 Fig. 55.—Attauapiskat, Ontario, summer, 1947. The Crce Indians of the James Bay region gather at such coastal posts as this soon after the breakup of the rivers in spring. In late summer or autumn they return to their inland trapping grounds. Those who trap far inland leave be- fore the autumn hunt for blue geese and snow geese begins along the coastal marshes. Fig. 56.—Summer scene at ,\ttauai)iskai. .\tter a luiij; '.\iiucr of arduous trapping and hunt- ing, often entailing considerable hardship, the native Indians are usually content to summer quietly at or near the coastal trading posts. lJi8 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 perhaps the barren-grounds type, migrated along the west coast of Hudson and James bays as far south as Fort Albany. They are now gone except for a small band at Cape Henrietta Maria, which may rep- resent remnants of this migratory group. Woodland caribou are found scattered over the muskeg country in small bands, but their total numbers are not great. When the long and dreary winter has fully expended itself, and the willow grouse (Tetrao Saliceti) have taken their depar- ture for more northern regions, there is frequently a period of dread starvation to many of the natives, who are generally at that time moving from their wintering grounds to the trading posts. The first note, therefore, of the large gray or Canada goose (Anser canadensis) is listened to with Fig. 57.—Through the establishment of a system of preserves and regulated trapping, beaver populations in the Hudson-James bay region are gradually being restored to former levels. Besides furnishing many pounds of highly nutritious meat and thereby reducing the hunting pressure by Indians on waterfowl, beavers also improve the character of small streams as brooding areas for Canada geese. This illustration shows a beaver dam on Little Partridge Creek, which empties into the southwest corner of James Bay. The tall trees that border the stream are black spruce; the principal shrubs are willow, alder, and sweet gale. Moose, always quite abundant in the country just south of James Bay, were scarce in the muskeg belt lying west of the bay until 1946 and 1947, when there was an unprecedented influx of these animals, presumably from the south and west. On the whole, however, except for the waterfowl he kills in spring and autumn, the James Bay Indian must rely on small game, such as snowshoe hares, muskrats, grouse, and ptarmigan for his meat supply. When these cyclic species fail he is usually in dire straits. Barnston (1862) wrote: a rapture known only to those who have endured great privations and gnawing hun- ger. The melancholy visages brighten, and the tents are filled with hope, to which joy soon succeeds, as the happy father, or the hopeful son and brother, returning success- ful from the hunt, throws down with satisfaction and pride the grateful load. Although the economic plight of the Indian has been gradually improved from those early times, particularly in re- cent years, through Dominion government family allowance, government relief, and liberal credit at the fur posts, the first I March, 1950 Hanson & Smith : Canada Geese 139 arrival of geese in the spring is always an event of great importance. Bishop Robert J. Renison of the Diocese of Moosonee relates one of the highlights of his early years as Anglican minister at Fort Al- bany. A funeral service had just been held at the small church and the mourners, cold, sick, discouraged, and hungry after a long winter, were moving on snowshoes toward the cemetery (Renison 1944). The Missionary walked in front, tread- ing warily among the tents where husky dogs prowled, on his way to the little grave yard where two men with pickaxes had been for hours chipping the frozen earth deep enough to make a shallow trench. Al- though in the morning the whole scene looked and felt like the ragged end of win- ter, now the South wind grows warmer every moment and already the haze is seen in quivering waves over the melting ice and snow. As the cortege was lost in the maze of wigwams, suddenly the cry of wild geese was heard. The funeral procession stood still and from all over the settlement came the answering call from every living soul. A great flock of Canada grey geese swept like a gigantic airplane over the trees re- joicing at what seemed a welcoming call. The phalanx turned to leeward and sailed slowly down over the spot from which the sounds came. It was too much even for sorrow and decorum. The Chief Mourner dived into his tent and appeared in a moment V. ith his loaded gun. With incredible ease and grace he brought down a goose with each barrel. Cheers and laughter rang out. The oldest instinct of man triumphed in every simple heart and as the pallbearer"; patted the bereaved husband on the back, he modestly replied like a true sportsman, "She did it. I always had luck when she was with me." Then the spell was broken ; the procession resumed its direction. The recent increase of beaver through restocking and the establishment uf bea\er preser\es on the west side of James Bay will, now that trapping is open, add thou- sands of pounds of highly nutritious meat to the Indian food resources, fig. 57. Since beaver and most of the Canada geese are secured in early spring, beaver restoration will materially reduce the annual toll of geese. This shift in hunting pressure is re- ported to have taken place in the Rupert House country where beaver trapping has recenth been on a sustained-vield basis. The spring kill of the Canada goose west of James Bay takes place inland when the Indians are still on their tiap- ping grounds and the rivers are frozen over. Hunting is done from blinds or stands built of brush and set out on the Fig. 58.—Decoys made by Cree Indians hunting in Hannah Bay at the south end of James Bay. The decoy in the top picture was made of willow twigs; the lifelike decoy in the lower picture was made from a log and a charred stick. river ice. Decoys made of willow twigs or small stumps or blocks of wood of proper size are set up in such a way as to bear a crude resemblance to a flock of sitting geese, fig. 58. Often using in- ferior arms with hand-loaded shells, the native hunters easily overcome the handi- caps of poor equipment by their expert ability to call geese, an art practiced from 140 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 childhood. In late summer, some of the native hunters with the aid of an inter- Indians supplement their meager diet by preter, fig. 60. In some cases it appeared hunting ducks along the coast, fig. 59, that the hunter questioned could remem- while in autumn most of the hunters are her his exact bag of the current year and in the coastal marshes for blue geese and of the previous year. In many other cases, Fig. 59. — Indian encampnnent on Cape Henrietta Maria. The Indians of this group trade at the Lake River outpost, but visit Attawapiskat briefly in the summer. Before autumn, they return to the cape to hunt waterfowl. snow geese, hoping to accumulate a supply of meat for at least a part of the winter. Any Canada geese killed at these times are incidental to the hunt for "wavies," as then the latter outnumber the Canadas along the coast by the ratio of many hundred to one. Our bag data were secured from post managers and other informed residents and through direct questioning of the it was equally obvious that the hunter could remember only the approximate number of geese killed and bagged, as he gave figures in multiples of 5 or 10. The inherent tendency to exaggerate in giving "rounded off" figures introduces con- siderable error. Therefore, we believe that the data in table 11 may exceed the actual bag by perhaps 10 to 15 per cent. A few Indians, fortunate enough to March, 1950 Hanson & Smith: Canada Geese 141 M'^ Fig. 60.—The number of Canada geese killed by Indians was calculated trom intorination secured through personal interviews with native hunters. The hunter being questioned here, with the aid of an interpreter, is a member of the Ogoki band. Sixteen hunters of this band were interviewed in 1946 and the same number the following year. The total number of Cree hunters interviewed was 94 in 1946 and 171 in 1947. Table 11.—Number of Cree Indian hunters, average bag per hunter, and total calculated bag of Canada geese by native hunters residing in the breeding range of the Mississippi Valley goose population, 1946 and 1947. Fur Trade Post and Indian Trapping Territory Ogoki Fort .'\lbany fincluding Kapis- kau and Ghost River out- posts) .Attawapiskat' (including Lake River outpost and .^kimiski Island) Wecnusk Fort Severn Total Average Total Number OF Hunters 16 100 134 33 47 330 Number of Hunters Interviewed 1946 16 24 28 26 94 1947 16 67 31 31 26 t7t Average Bag per Hunter Interviewed 1946 3.0 9.S 13.3 1S.0« 14.0 65.0 13.0 1947 3.6 11.1 15.6 19.0 17.0 66.3 13.1 Calculated Bag per Trapping TERRiToay 1946 48 950 1,782 495 658 3,933 1947 56 1,110 2,090 627 799 4,68S 'The bag at AttawapUkat in 1948 was 1.720 according to Dr. John Honigman. resident anthropologist at the post that year (personal communication). ' An estimate, based on data for later year. 142 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 ha\c their trapping grounds located in good Canada goose hunting territories, bag as many as 45 geese per hunter, while other Indians, located in poor goose habitat, take only a few geese or rione. In 1944, when the inventory showed that there were approximately 66,000 geese in the Mississippi fl\way, the estimated bag on the breeding grounds was 5,500, or about 8 per cent of the number of birds believed to have been available to the Indians in the spring of that year. In 1946 and 1947, the calculated bag, table 11, represented about 10 and 9 per cent, respectively, of the total population avail- able in the springs of those years, table 7. Band recoveries, on the other hand. indicate that the annual bag of the na- tives is about 5 to 6 per cent of the avail- able population. Taking into considera- tion the kind of error inherent in these data, it would seem that the Indians do not kill more than 10 per cent of the Canada goose population that reaches the breeding grounds in the spring. The time of kill of Canada geese by Indians on the breeding grounds is in- dicated in fig. 61. Southern Canada and United States Available data on the goose bag in the United States and in southern Canada are very unsatisfactory. Goose bag records from Illinois are more nearly complete 60-1 GO 50- 40- > O o UJ tC 30- li- o 20- 3 _; 10- APR. MAY JUNE JULY ¥ AUG. SEPT. OCT. ::s NOV. MONTH OF RECOVERY Fig. 61.—Time of kill of Canada geese by Indians on the breeding grounds, as shown by re- covery records, 1941-19-f7, of geese banded at the Horseshoe Lake Game Refuge. March, 1950 Hanson & Smith: Canada Gkese 143 Table 12.—Estimated ba^ of Canada geese in regions frequented by the Mississippi Valley population, 1941-1945. Province, State, or Other Area West coast of James B Ontario, exclusive of James Bay region Minnesota Wisconsin Michigan Indiana Illinois OhI Iowa Missouri Arkansas Kentucky Tennessee Mississippi Louisiana (eastern p.irt onlyj Total Estimated Mean Bag 4,700 400 200 620 000 500 8^400 200 75 260 300 50 100 400 150 19,355 than those from any other state, in part because of a law that requires all licenced clubs to keep daily records of their take and in part because only a small number of areas in the state afford goose hunting. Most states in the fl\-\vay, however, at- tempt to calculate the take, either from report cards attached to the hunting li- censes ( the cards are designed to be mailed to a state official at the end of the season) or from data obtained from questionnaires sent to a sample of the licensed hunters. A few states make no attempt to secure bag data. Bag data based on the hunter-report- card system are apt to be e.xaggerated. Studies made by Bellrose (1947) ha\e shown that the state-wide bag of ducks in Illinois calculated from report cards is several times the actual bag. In AViscon- sin, the calculated bag of Canada geese for two counties has been from 3.5 to 4.6 times the actual bag (see section on Wis- consin, below). Furthermore, as many states do not record the goose bag b\' species, the actual portion of the calculated bag that consists of Canada geese can only be estimated. Table 12 summarizes our information on the Canada goose bag in recent years in those regions that lie in the Mississippi flyway. A more detailed anal\sis of the bag follows. Southern Ontario.— In spite of the fact that the Miner Sanctuary, at Kings- ville, in Essex County, has been a heavy concentration point for Canada geese for over 20 years, no commercial shooting clubs have operated in the fields sur- rounding the sanctuary. All of the hunting in the Kingsville area is re- ported to be flight shooting from public roads as the geese go to and from the sanctuar\' and their roosting grounds on Lake Erie, The number of banded geese reported taken in this area does not indicate the true size of the bag since some of the local shooters do not appreciate the im- portance of the banding program at the Miner Sanctuary and do not report the bands they recover. When live decoys and baited fields were permitted, the autumn bag in Essex- County was about 1,000 birds, but, since these practices were outlawed, the bag has probably not exceeded 500 and fre- quentlv is as low as 200 or 300 birds. We are informed that the 1945 kill was unusually low, not over 50 geese. We are told that near the Miner Sanc- tuary it is possible to bag geese easily only on days when there is a heavy overcast and a strong wind is blowing, thus causing the ge«se to fly low. On most days the geese are reported to be well out of gun range when they pass over the hunters who shoot on the perimeter of the protected area, the radius of which extends 1 mile beyond the sanctuary property. 144 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Table 13.—Recoveries in the Mississippi flyway,* 1925-1944, of Canada geese banded each autumn and winter at Kingsville, Ontario. March, 1950 Hansox & Smith: Canada Geese 145 Table 15.—Annual shooting losses of Canada geese in the region of Horseshoe Lake, 1941-1945. Hunting Season 146 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 cording to H. J. Miller of the Michigan Department of Conservation. The figures given in table 16 are not comparable from year to year, because between 1938 and 1944 the number of licensed hunters sending in bag reports to the Department of Conservation dropped from 65 to 20 per cent. Because success- ful hunters are more apt to turn in their report cards than unsuccessful hunters, the accuracy of the computed kill varies from year to year. Excluding from table 16 the probable bag of blue geese and snow geese and taking into consideration exaggeration inherent in calculating state- wide bags from hunter report cards, we estimate that the annual kill of Canada geese in Michigan from 1938 through 1944 was between 1,000 and 3,000 birds. In 1945, the total calculated bag of all species of geese was more than 23,000. Of this number, well over half consisted of blue geese and snow geese (Dr. Miles D. Pirnie of Michigan State College, per- sonal communication) that failed to make their usual rapid southward migration and that were observed and shot in un- usual numbers. The large number of banded geese reported shot in Michigan in the autumn of 1945 does, however, fur- nish undeniable evidence that there was a large increase in the total bag of Canada- geese in that year over the number bagged between 1938 and 1944. Of 20 Canada geese banded at Horseshoe Lake between 1940-41 and 1944-45 and bagged in Michigan, table 14, 13 were bagged in the autumn of 1945. The following Michigan counties, which are in the vicinity of important autumn concentration points and winter- ing areas, yielded the largest bags of geese of all species: Chippewa County in the northern peninsula; Leelanau County in the northwest sector of the lower peninsu- la; Huron, Tuscola, and Bay counties bordering Saginaw Bay; and Allegan, Kalamazoo, Barry, Berrien, and Calhoun counties in the southwestern lake section. According to Dr. Miles D. Pirnie of Michigan State College (personal com- munication in 1945) between 500 and 1,000 Canada geese were bagged within a 20-mile radius of the W. K. Kellogg Refuge, Barrv and Kalamazoo counties, in 1944. Various counties represented by 159 banded Canada geese shot in Michigan, 1925-1944, are shown in table 17. Table 17.—Recoveries of Canada geese in Michigan, 1925-1944, banded at Horseshoe Lake, Illinois, and at the Jack Miner Bird Sanctuary, Kingsville, Ontario. COUN Allegan . . . . Berrien Barry Huron Kalam.izoo. . Calhoun. . . . Sanilac Monroe. . . . St. Joseph. . Van Buren . Chippewa. . . St. Clair. . . . Mackinac. . . Cas.s Washtenaw. Wayne Newaygo . . . Number OF Ri COVERIES 31 16 13 12 12 11 10 5 5 4 4 4 3 3 2 2 2 Total recoveries. CouN Presque Isle. . . Alpena Benzie Lenawee Livingston. . . . Otsego Gratiot Gladwin GrandTraver.se Saginaw Roscommon. . Luce Kent Oscoda Lapeer Eaton Lake Number OF Re- coveries 150 Whatever the actual bag of Canada geese is in a given year in Michigan, it probably is not all at the expense of the Mississippi flyway population, as some geese belonging to the Southeast popula- tion undoubtedly are bagged as they migrate down the eastern edge of the state. Wisconsin.—Band recoveries, table 14, indicate that Wisconsin is second only to Illinois in the toll its hunters take of the Horseshoe Lake flock. The largest kills of Canada geese in Wisconsin are made in Rock and Walworth counties, in the vicinity of the Rock County Refuge. Geese in the Rock-Walworth county area show little of the lameness exhibited by the Horseshoe Lake flock. On leaving the refuge on the upland prairie for their roost lakes, they are said generally to spiral high up out of gun range before crossing over its boundaries, thus account- ing in part for the relatively small kill, which is equivalent to about 8 per cent of the geese that are present in these two counties in late autumn and winter. March, 1950 Hanson & Smith: Canada Geese ;47 Careful estimates of annual bags in the Rock County Refuge area by personnel of tlie Wisconsin Conservation Department and the United States Fish and Wildlife Service, 1940-1945, did not exceed 4U0 birds; the average annual bag was con- siderably less. ^Visconsin Conservation Department estimates for the probable and maximum Canada goose bags in these counties are as follows: 1940—200 1941—75, not exceeding 150 1942—317, not exceeding 400 1943—150, not exceeding 20U 1944—40, not exceeding 50 1945—350, not exceeding 400 When the above estimates are compared with the bags calculated from hunters' report cards (1940, 732; 1941, 581; 1942, 1,445; 1943, 629), it appears that the annual calculated bags are exaggerated 3.5 to 4.6 times. These calculated bags are derived from a sample of the kill cards sent in by about 35 per cent of the licensed hunters. If the calculated state- wide bags reported by the Wisconsin Conservation Department are exagger- ated to the same degree as are bags for Rock and ^Valworth counties (3.5 times), the corrected state-wide annual bag of Canada geese in Wisconsin between 1932 and 1944 has averaged about 500 birds and varied from about 170 (1935) to 860 (1942). If our method of estimating the state-wide bag is sound, it appears that the annual kill of Canada geese in Wisconsin has seldom approached the thousand mark. Important kills have also been reported for Waushara County. The bag in this county in 1942 was estimated by Zimmer- man (1942) to be 400. The total num- ber of migrant geese that offered shooting to hunters in this area is unknown. Minnesota.—Most of the Canada goose hunting in Minnesota is said to occur in the western third of the state, especially during the wet years. Kills in the eastern sections rarely occur, so that the total bag of the Mississippi flyway geese in Minnesota is probably small. The lack of band recoveries from eastern Minnesota substantiates this belief. From 1935 through 1944, the computed state-wide bags of all species of geese, based on re- ports received from 10 per cent of the hunters, ranged from 1,869 to 5,050 birds. As in Michigan, the 1945 calculated bag., for all geese was the largest on record,,, 10,908. Ohio.—We have few- kill data for Ohio otiier than band recoveries. The princi- pal kills of importance to Mississippi fly- way geese would be those made in the region of Lake St. Marys. Kills made in central and eastern Ohio would be pri- marily at the expense of the Southeast fly- way geese. We have arbitrarily placed the bag of Mississippi flyway Canada geese in Ohio at 200 per annum. Indiana.—According to William B. Barnes of the Indiana Department of Conservation, goose hunting in Indiana is heaviest in the Kankakee region of northwestern Indiana and in the lake dis- trict to the east. As the flights move through northern Indiana to the south- west, additional shooting is provided in the Wabash River valley. Hunting pres- sure in this state appears to be, on the whole, relatively moderate. Of the total number of recoveries of geese banded at the Miner Sanctuary during the autumn in the past 20 years, approximately 8 per cent have been from Indiana, table 13. At Hovey Lake Refuge about 300 Canada geese are generally present during the open hunting period, and the largest bag in any one season in a 5-year period, 1940—1944, was only five birds, 1.6 per cent of the flock. Partly responsible for this small bag was the wildness exhibited by the geese in the refuge vicinity. Judging from questionnaire answers received from hunters by the Indiana Department of Conservation, it is doubt- ful if the kill of Canada geese in Indiana in recent years has ever greatly exceeded 2,000 birds and probably in most years the kill is considerably less than this figure. Iowa.—According to Bruce F. Stiles of the Iowa State Conservation Commis- sion, the yearly kill of Canada geese in Iowa is about 1,200 birds. He states that the heaviest migration is down the Mis- souri River valley. As band recoveries indicate that central and western Iowa is well west of the migration routes of the Mississippi fl\way population, only a small portion of the above kill would be at the expense of this population. The paucity of band recoveries from eastern Iowa, 148 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 tables 13 and 14 and figs. 13-21, signifies that few Mississippi flyway geese mi- grating through this sector of the state stop en route long enough to afford much shoot- ing. Missouri.—The Missouri Conserva- tion Commission estimates that, prior to the establishment of the Horseshoe Lake Game Refuge, approximately 75,000 Canada geese wintered on the sand bars and islands of the Mississippi River be- tween Ste. Genevieve and Caruthersville, Missouri. Band recoveries, table 13, with the exception of returns from 1935 through 1939, indicate no pronounced change in the Missouri kills in relation to the Illinois kills since 1925. Before 1941, when the geese using the Horseshoe Lake Game Refuge were reported to have made daily flights to the river bars, considerably larger kills are said to have been made in Cape Girardeau, Scott, and Mississippi counties than in more recent years. Band- recovery data indicate that this period of higher kills was between 1935 and 1939. The yearly bags, estimated for the above counties by Paul S. Smith, are given in table 15. According to information re- ceived from M. O. Steen of the Missouri Conservation Commission, the annual bags in Missouri in the region of Cape Girardeau, Scott, and Mississippi counties averaged approximately 175 geese in re- cent hunting years. State-wide annual bags, 1943—1945, are estimated to have been less than 400 birds. Besides the bag in southeastern Missouri, about 125 geese were killed on the Missouri River between Booneville and Jefferson City in central Missouri, and approximately 100 were killed in the vicinity of Swan Lake National Refuge in the north central part of the state. However, on the basis of present evidence, it would appear that the geese killed in central Missouri belong to the Eastern Prairie population and are not Mississippi flyway birds. Considerable numbers of Canada geese are reported to migrate through central and southwestern Mis- souri in the autumn, and it seems reason- able to conclude that they winter in west- ern Louisiana and eastern Texas. Kentucky.— Little information is available in regard to the state-wide kill in Kentucky, though band recoveries indi- cate that only the kills made in the western portion of the state would be from the Mississippi Valley population. Band re- coveries show that since 1940 the annual bag of Mississippi Valley geese in this state has been greatly reduced, table 13. In 1939 and 1940, Paul S. Smith estimated that about 100 geese from the Horseshoe Lake flock were bagged in Kentucky; in more recent years, band recoveries and the findings of reliable observers indicate that very few geese from the Horseshoe Lake flock have been shot in Kentucky. Tennessee and Mississippi.—The section of the Mississippi River bordering Tennessee, Arkansas, and Mississippi may be considered as a single unit insofar as the kill of Canada geese using the river bars is concerned. In 1943, it was esti- mated that not over 50 geese were killed on and in the vicinity of the Tennessee section. It is the belief of W. F. Dearman, formerly director of the Mis- sissippi Department of Fish and Game, that the 1943 bag for his state along the Mississippi River was approximately 800. Arkansas.—Kills of Canada geese in eastern Arkansas, exclusive of the Mis- sissippi River, are made over such an ex- tensive area and in such relatively small numbers in any given locality that it is difficult to make an accurate appraisal of the over-all loss. In 1943, the bag was about 400, and in 1945 it was probably even lower. After talking with hunters, employees of hunting clubs, and em- ployees of local cold-storage plants, we concluded that the bag of Canada geese in the Stuttgart region in 1945 did not exceed 200. Louisiana.—In 1943, losses of Canada geese through hunting in the delta and coastal marshes were estimated to be approximately 1,000. Of this number about 150 were estimated to be Mis- sissippi flyway geese; the greater portion of the Canada goose population in Louisi- ana is in the western portion of the state and probably belongs to the Eastern Prairie population. Total Annual Bag Before sound management measures can be instituted for the Mississippi Valley Canada geese, the over-all kill in the population must be known within March, 1950 Hanson & Smith: Canada Geese 149 fairly close limits. We do not have com- plete data on the kill, but a reasonably accurate appraisal can be made from avail- able information. Table 12, summarizing bag data con- tained in previous discussions, is fairly accurate in some instances and in others represents very rough estimation. It should be remembered that the lowest and highest bags for the various areas repre- sented did not occur in the same calendar year; hence, totals for those respective columns do not represent annual extremes. It would appear from table 12 that the average annual bag in the flyway, 1941— 1945, was somewhere in the neighborhood of 19,000 birds. The annual loss of geese through hunt- ing, expressed as a percentage of the population that left the breeding grounds in the autumn, may be roughly estimated* for the Horseshoe Lake flock and the fly- way population as a whole. The number of geese calculated to escape death from natural causes after leaving the breeding grounds and to be subjected to hunters' guns may be arrived at by adding known hunting losses to inventory figures after the hunting season. For example, the Horseshoe Lake flock num- bered about 37,000 geese at the time of the 1943—14 inventory, table 9. Local losses in the Horseshoe Lake area, in- cluding crippling, were approximately 16,000 geese, table 9. Assuming that losses between the Canadian border and Horseshoe Lake were average that year, an additional 3,250 geese (2,600, a figure based on band recoveries, plus an assumed 25 per cent crippling rate) were lost. The autumn kill by the Indians on the breeding grounds is small, fig. 61, as is also the kill by white hunters in southern Canada. Including crippling losses of 25 per cent, the combined kill may be in the neighborhood of 800 birds, about half of which would be contributed by potential Horseshoe Lake geese. Of the 37,000 geese leaving Horseshoe Lake in the spring, approximately 8 per cent are bagged by the natives plus an estimated additional 2 per cent lost through crip- pling, or a total of 3,700 geese lost. The • Accur.icy of the following estimstions is in Urge measure dependent on the accuracy of inventory figures used in the computations. estimated combined total of all hunting losses for 1943-44 was 23,350. Inventory figures plus hunting losses for 1943—1-4 (omitting the spring Indian kill which occurs after the inventory) in- dicate that the Horseshoe Lake popula- tion that left the breeding grounds in the autumn of 1943, and subsequentl.\' eluded death from other causes during the follow- ing 6 to 8 months' period, was roughly 56,650. Thus, total losses through hunt- ing (including spring losses in Canada) are computed to have been about 41 per cent of the geese that survived death from natural causes. When crippling losses are deducted, it appears that hunters bagged about 30 per cent of the geese that sur- vived death from natural causes. Over-all loss rates due to hunting for 1944-45 and 1945-46, calculated in a similar manner, were approximately 39 and 40 per cent, respectively, of the population that survived other types of mortality. Hunting losses for the flyway popula- tion as a whole, as might be expected, were at a considerably lower rate than for the Horseshoe Lake flock. In some recent years, the bag of geese in the flyway has been about 19,000, table 12. In some of the same years, inventory figures, table 7, indicate an average population of ap- proximately 60,000. If the bag prior to inventory (roughly 14,600) and the over- all crippling, arbitrarily placed at 25 per cent (total 18,250), are added to the ap- proximately 60,000 surviving at inven- tory, an original population of 78,250 is indicated. Thus, of all flyway geese that survived natural mortality during recent hunting periods, at least 23 per cent are estimated to have succumbed to hunters. Canada vs. LJnited States Kill Are the people of Canada, especially the Indians and Eskimos, getting an un- justifiably large share of the Mississippi flyway Canada goose population ? Many hunters in the United States would like to believe that such is the case. However, investigators (Soper 1930, Sutton 1932, Brandt 1943, Gillham 1948) of bird life in the far north believe that in most in- stances the future of waterfowl popula- tions in arctic and subarctic regions is not threatened by the kills made by the native 150 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 peoples. It is their belief that the fate of waterfowl populations breeding in the north will be decided by the treatment ac- corded them on their wintering grounds. We likewise believe that the future of the Mississippi Valley population is dependent on the protection and care it is given south of the breeding range. On the breeding grounds of the Mis- sissippi Valley Canada geese, there has been a decrease, in recent years, in the number of Indians dependent upon the game resources of the country. After World War I, many of the Fort Albany Indians moved to new trapping grounds far into the interior. According to Dr. T. J. Orford, formerly Indian agent at Moose Factory, in 1945 there were 124 Indians from the Fort Albany band at Lac Seul, a locality to which they had moved in the 1920's. There was another exodus of Fort Albany Indians from the James Bay area in 1942 when 150 transferred to the Constance Lake band on the Canadian National Railway line. Additional Indian families moved down to Moosonee from Fort Albany and Attawapiskat during the years of World War II. As a result of these movements, Indian hunting pressure on wildlife in the James Bay area has decreased. In contrast, the number of hunters shooting Canada geese in the United States, notably in Illinois, has in- creased tremendously since World War I. Data in table 12 show that the take in Canada, 1941-1945, was roughly 25 per cent of the total bag of Mississippi Valley geese. When it is remembered that the Indians are partly dependent on geese for survival, that their kill is not a new drain on the goose population, and that in recent years the kill has been found to be pro- portional to the goose population, this kill cannot be considered excessive. The relative kill by the Indians and by hunters outside the breeding grounds in Canada and in the United States can be found by comparing the number of geese killed by each group to the total number of birds available to each. It was shown earlier that the Indians kill about 10 per cent or less of the goose population avail- able to them. As the kill in Canada away from the breeding grounds is estimated as not exceeding 1 or 2 per cent of the population, the total Canadian kill is concluded to be 10 to 12 per cent of the population available in any year. A rough measure of the bag contributed by the flock between the Canadian border and the Horseshoe Lake Game Refuge can be derived from an analysis of band re- coveries. Between 1941 and 1944, the number of band recoveries from north of the refuge was equivalent to 33 per cent of the number of recoveries in the region of Horseshoe Lake 1 or more years after banding. The number of geese to termi- nate their migrations at Horseshoe Lake, 1941-1944, averaged about 45,000. The known bag by licensed clubs in Alexander County in those years averaged 7,780, table 15. Figures based on estimates from band recoveries in 1941—1944 indicate that the flock contributed an average yearly bag of about 2,600 birds before reaching the refuge, or a loss of about 5 to 6 per cent of the numbers that crossed the Canadian border. As the Horseshoe Lake flock in recent years has comprised about 50 per cent of the Mississippi flyway population, the bag of Horseshoe Lake geese (2,600) computed from band recoveries for areas between the Canadian border and the refuge should, if doubled (5,200),* approximately equal the bag of all Mis- sissippi flyway geese in the same area. Calculations from data in table 12 indicate that the estimated mean annual bag for states in the flyway north of Horseshoe Lake (Minnesota, Wisconsin, Michigan, Ohio, Indiana, Illinois, t and Iowa) for 1941-1945 is 5,695, a figure close to the bag figure approximated from band re- coveries and from inventory figures and bag data from the Horseshoe Lake area. A check of band-recovery records for the period 1925-1944 shows that most geese bagged in this area were killed in Novem- ber, fig. 62. In the fall of 1943, when hunters between the Horseshoe Lake region and the Canadian border bagged between 5 and 6 per cent of the Mississippi flyway geese * This figure, based on a comparison of band-recovery rates, may be low for two reasons: (1) the percentage of hunters reporting bands ihey recover is probably lower over most of the flyway than it is at Horse-hoe Lake, where the importance of reporting bands has been well publicized, and (2) the geese that spend the greater part of the hunting sea on north of the refuge are subject to heavier shooting pressure in that region than are the Horseshoe Lake gee-e in the short time they are there._ t The figure for Illinois (about 1,100) does not in- clude the bag for the Horseshoe Lake area. March, 1950 Hanson & Smith: Canada Geese LSI available to them, hunters in the region of Horseshoe Lake bagged 23 per cent of the total number of geese attaining the refuge in the fall and winter of 1943-44. Figures for the Horseshoe Lake region, as calculated from data in tables 9 and 15, are 23 per cent for 1943-44, 19 per cent for 1944-45, and 18 per cent for 1945-46. It is desirable at this point to discuss a type of rumor at times common among waterfowl hunters. During the 1944 hunting season, several hunters at Horse- shoe Lake expressed the opinion that, if the ducks and geese needed further protec- tion, the Indians in Canada should be prohibited from gathering and selling duck and goose eggs to a company manufactur- ing pancake flour. 1 hat this kind of com- plaint is an old stor)" and has no basis in fact was shown by Grinnell (1901). Various explanations of the change [waterfowl decrease] are given. The blame is laid on the market shooter, on the supposed destruction of birds and eggs on the northern breeding grounds, and on supposed changes in the lines of flight by nu'grating birds, but most gunners are un- willing to accept the logic of events and to acknowledge that the principal cause of the lessened number of the fowl lies with the gunners themselves, and is an inevitable accompaniment of civilization, not to be changed except by radical measures One of the most grotesquely fantastic explanations of the scarcity of wildfowl was put forth several years ago in the news- papers: . . . This story told of an enormous destruction of wildfowl eggs in the North- west for commercial purposes; millions of shiploads and trainloads of such eggs, it was gravely related, being annually gathered in Alaska and British America, and shipped 10-20 21-31 OCTOBER -10 11-20 21-30 NOVEMBER 10 11-20 21-31 DECEMBER -10 JAN. Fig. 62.—Time of kill of Canada geese in Wisconsin, Michigan, and Indiana, as shown by recovery records of Canada geese banded at the Horseshoe Lake Game Refuge and the Jack .Miner Bird Sanctuary and recovered in the period 1925-1944. Migration dates and the time of most hunting seasons combined to make November the month of heaviest kill. 152 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 thence to points in the East, where they were manufactured into egg albumen cake. . . . This, then, was the conclusion of the whole matter: Those who professed to have information on the subject were unable to substantiate the stories which they told; the transportation companies have carried no such eggs; none have ever been received at the ports of entry; the albumen trade knows nothing whatever about them, and in view of the total lack of evidence to support the story, there is no doubt that it is a pure invention. DIFFERENTIAL HUNTING LOSSES To manage a wildlife species that is subjected each year to heavy gun pressure, it is important to know not only how many individuals of a population are shot annually, but also if the kill in each of the various age and sex groups is propor- tional to its size in the group, and if the kill places an undue burden on any partic- ular component. One of the causes of concern relative to Canada goose shooting at Horseshoe Lake in recent years has been the disproportionately large kill of juvenile birds, table 23. What are the underlying factors re- sponsible for a differentially heavier kill of the younger geese? One factor has already been mentioned, namely, the strong bonds existing between members of family units. Related factors are the fearlessness of j'oung geese and their dependence on adults for guidance during their first year of life. The relationship of juvenile age to un- wary behavior in Canada geese was in- BEFORE SEASON - 1165 OBSERVATIONS 50- ^ DURING SEASON - 278 OBSERVATIONS AFTER SEASON - 699 OBSERVATIONS O O UJ o a: UJ Q. ONE TWO NUMBER IN FLOCK THREE OR MORE Fig. 63.—Percentages of each of three size-groups (single bird, pair, group of three or more geese) in the total number of flock formations of Canada geese observed before the sea son, during the season, and after the season at Horseshoe Lake, 1945. March, 1950 Haxson & Smith: Canada Geese 153 NUMBER IN FLOCK NUMBER IN FLOCK HORSESHOE LAKE JANUARY 3, 1945 NO. OF FLOCKS, 191 40 30 20 10 23456789 JAMES BAY SEPTEMBER, 1944 NO OF FLOCKS, 48 40 30 20- 10- 23456789 NUMBER IN FLOCK Fig. 64.—Frequency counts of flocks numbering nine Canada goose flocks at four different locations. ROCK COUNTY, WISCONSIN NOVEMBER, 1945 .NQ OF FLOCKS, 298 , AV^RA9^ , , , .2345 6789 NUMBER IN FLOCK or fewer geese and the average size of timated by Phillips (1921). "It was re- marked by Massachusetts gunners that there seemed to be a large proportion of young geese, and the same was true of Currituck Sound, N. C, where geese also appeared in unusual numbers and were very tame. The tameness of the geese in Massachusetts this past season caused comment everywhere, and I saw instances of it myself." At Horseshoe Lake it was found that, as the shooting season progressed, the rel- ative number of single birds increased because of the breaking up of family units, figs. 63 and b4F, G. Single juvenile birds separated from their families were fre- quently observed to associate and feed with other family units, as well as with unattached adults, but they were often at the bottom of the peck order, and, as they never appeared to be accepted into the ranks of other families, they often flew alone. Every veteran goose hunter knows that single birds are "suckers," more readily decoyed than pairs or flocks. During the 1945 season, Arthur S. Hawkins, then with the Natural History Survey, and the 154 Illinois N.atur.al History Survey Bulletin Vol. 25, Art. 3 Table 18.—Number of Canada geese shot and size of flock from which they came. Observations at Horseshoe Lake in 1945. Number IN Flock Contrib- uting to Kill .March, 1950 Hanson & Smith : Canada Geese 155 Our calculations indicate that the aver- age yearling was about 2.8 times as vulner- able to hunting by the Indians on the breeding grounds as was the average adult (older than 1 \ear) in the springs of 1942, IQ43, and 1^44. Since the bulk of the Indian kill is made from .April 15 to June 1, fig. 61, 5 months after the close of hunting in southern Illinois, it would seem that experience gained by the young geese in that interval does not greatly reduce their vulnerability on the breeding grounds. A crucial period that immediately follows their abandon- ment by the adults results in continued vulnerability of the yearlings to gun pres- sure. There is ample evidence tiiat this aban- donment occurs just prior to nesting. L'n- less broken up by shooting, family groups at Horseshoe Lake are often maintained throughout the autumn and winter period. From observations made at his sanctuary, Jack Miner (1923) believed that goose families do not break up until they reach the breeding grounds. This belief is sub- stantiated by the Indians, who have ob- served that the young of the previous year are separated from the adults shortly be- fore the breeding season. The breeding adults in the captive goose flock at the Bright Land Farm near Harrington, Illinois, according to Charles Kossack of Harrington, are similarly known to drive off their yearling \oung at nesting time. "Cast oft" young geese are on their own, without the guidance of adults, and prob- ably are associated at first in small groups. One of the reasons for the differential vulnerability of the yearlings on the breeding grounds was suggested in 1946 by John Gunnar and Gilbert Faries, life- long residents in the James Bay area and experienced goose hunters. They stated that, when the first geese arrive in the spring, many of them are very tame and curious. Gunnar volunteered the opinion that the first arrivals are the nonbreeding geese (yearlings). He recalled that on one occasion, while he was in the Part- ridge Creek area, a flock of inquisitive Canada geese decoyed within 10 feet of his head, and he expressed the belief that the white garment he wore at trc time was responsible for tlieir curious be- havior. Similar lack of wariness in the other species of geese has been noted. Brandt (1943) says of the white-fronted goose at Hooper Bay, Alaska: "And immediately after the lifting of the ice embargo these groups disintegrated into mated pairs, excepting small bunches of bachelor males. These free-lance gallants, often in com- pany with like possibly rejected suitors of other species of geese, spend their time moving abstractedly around in inquisitive flocks, and are ludicrously easy to decoy."* Of the blue goose, Soper ( 1930) writes: "With the breeding birds resuming, or commencing nesting duties large numbers of nonbreeding geese were left to fly aimlessly about in carefree existence dur- ing the brief span of the arctic summer. These were the restless and irresponsible flocks and individuals which from now on were to be obserxed in the Camp Kungovik locality." Tlie liigher mobility of \earling geese as compared with that of nesting pairs is a factor that may make the xoung birds readily available to Indians. Nesting adults are known to be extremely wary and secretive, and, unless they are especial- ly sought after, their presence may be known only by chance. Lack of wariness on the part of yearling geese apparently lasts until they begin to band together. In the summer, large flocks, believed to be comprised mainly of yearlings, are ex- tremelv wary. CRIPPLING LOSSES As crippling losses are a component of hunting mortality, they should be con- sidered a part of the total yearly allow- able kill in any game species. Whether a goose ends up on the hunter's table or dies of wounds and furnishes a banquet for some scavenging predator, the net loss to the flock is the same. Reduction of the crippling loss must always be an objective if maximum utilization of a game species is to be achieved. For many years, waterfowl shooting has been known to produce a considerable loss of unretrieved cripples—a loss that is high • Ii.ilic^ by llie autlior^ of this p.iper. Boardman Con- over, w\\n was in the Hooper Bay area with Brandt, has informed the authors that many flocks exhibiting this kind of behavior were composed of nonbreeding yearlings. 156 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 in proportion to the number of birds bagged. In most cases the crippling loss reported for ducks amounts to at least 30 per cent of the number of birds bagged and in some situations one duck is lost for cverv one bagged (Errington & Bennett 1933, Hawkins & Bellrose 1939, Baum- gartner 1942, Hochbaum 1944). In Michigan about 10 per cent of 105 ducks trapped and examined carried shot (Whit- lock & Miller 1947), but what percent- age of these ducks died later as a result of the prolonged effect of carrj'ing shot is not known. Recent fluoroscopic studies made by the Illinois Natural History Sur- vey of ducks trapped at Spring Lake on the Mississippi River and at Lake Chau- tauqua on the Illinois River revealed that approximately 25 per cent of the mallards migrating through these areas carry lead shot in their bodies as a result of shoot- ing. Goose shooting at Horseshoe Lake, 1940—1945, resulted in crippling losses similar to those reported to occur in duck hunting. To anyone who observed the shooting at clubs bordering the refuge at Horseshoe Lake in the years of this study, it was apparent that the height at which a goose flew over the hunters seldom deter- mined whether it was shot at. The situa- tion was aggravated by the heavy concen- tration of hunters ; hunters in the first line of pits or blinds attempted to "reach" approaching geese before the birds flew over the next line of pits. Novice goose hunters usually underestimated distances, while expert shooters, disgusted with the ease with which geese leaving the refuge could be killed, sometimes found sport in attempting to "scratch down" the high birds. High shooting, some observers believed, saved large numbers of geese by frighten- ing them off before they could fly within killing range. This was undoubtedly true during the early part of a season when the geese were not working out of the refuge in great numbers, or in years when low kills were made, but late in a season when geese were so numerous in flight over club grounds that the majority of hunters, even those who indulged in high shooting, got their limits, or in a year of high kill rate when the season was limited by a pre- determined kill, high or indiscriminate shooting was a factor certain to cause needless crippling and increase the total loss. In 1944, a questionnaire was circulated among goose hunters to obtain their own appraisal of their shooting. During the 21 -day season, 103 hunters were asked questions about the following items : num- ber of shells fired, estimate of geese light- ly hit, number of geese severely crippled and not retrievable, and the number of geese bagged. An analysis of the ac- cumulated data shows that the average bag per hunter-day was 1.69 geese. Since the average hunter success for all clubs in the vicinity for the entire season was 1.44 geese per hunter-day, it can be as- sumed that a fairly representative group of hunters was sampled. The 103 hunters reporting estimated that with 1,374 shells they had bagged 286 geese and had severely crippled 51 geese ; the number of geese crippled was equivalent to 18 per cent of the number bagged. This percentage probably repre- sents the minimum crippling loss. The hunters reported that they had lightly hit an additional 176 birds, or a number equivalent to 61 per cent of the number bagged. Thus, according to their own estimates made the day of hunting or a day after, these hunters hit, with varying degrees of severity, and did not recover, a number of geese equaling 79 per cent of the number that they recovered. How- ever, this figure is so high as to cast some doubt on its validity. In 1945, Arthur S. Hawkins, then of the Illinois Natural History Survey, and the authors observed the shooting at several clubs and made on-the-spot tallies of the number of geese bagged and the number crippled but not recovered. The tally of crippled birds included only those that had been obviously and severely hit, but others may have suffered mortal body wounds without exhibiting a noticeable reaction to their wounds at the time of being shot. The hunters under observa- tion bagged 253 geese but failed to re- cover an additional 62 badly crippled birds, most of them able to fly well enough to regain the lake within the refuge bound- aryr, but so severely crippled as to be un- able to survive the winter. Thus, in addition to each four geese bagged, ap- March, 1950 Hanson & S.^rITH : Canada Geese 157 proximately one additional goose died as a result of shooting—a minimum crippling loss of 25 per cent. At a few clubs the ratio of birds crippled to birds bagged frequently exceeded a ratio of one to one. Two instances of e.xtreme crippling were observed : in one, four geese were crippled and none bagged, and, in the other, seven were crippled and six bagged. From the various data presented above, we conclude that a conservative figure for the over-all loss owing to crippling at Horseshoe Lake is at least 30 per cent of the total bag. Crippling data are lack- ing from other areas in the flyway, but it is doubtful if the rate attained at Horse- shoe Lake was exceeded. Where shooters are widely spaced and are not competing with each other to knock down the same high-flying birds, there is relatively less wild firing and hence less crippling. Since crippling is more or less directly related to the number of shells fired, in- formation was sought on the number of shells the average hunter expended to secure one goose. The hunters canvassed by questionnaire in 19-H- reported that they fired an average of 4.8 shells per goose bagged. In 1945, data of a similar nature were obtained by an examination of shooting pits at the end of the first day of hunting. Of the 42 pits examined at two club shooting grounds bordering on Horseshoe lake, the average pit contained 37 recenth- fired shell casings. As each of the hunt- ers at these clubs killed his limit of two geese, and as no more than two hunters were permitted in each pit, the average number of shells fired to kill one goose on opening day in 1945 was nine. How does this score at Horseshoe Lake, where goose shooting was relatively easy, compare with goose-shooting scores else- where? On the basis of his goose-hunting experiences in the West, Major Askins (1945), a noted authority on arms, be- lieves that one goose to three shells, when distances are less than 80 yards, is about the best score an average hunter can ex- pect. Most of his shooting was of the pass variety, and he states that it is doubtful if his score was better than one bird in four shots. Geese wounded near Horseshoe Lake generally attempted to regain the lake either by flying or by eluding the hunter on the ground. Since hunters were not permitted to recover cripples that entered the refuge, club owners, at the opening ot the 1943 season, were required to erect a 2-foot woven-wire fence between the pits and the lake to aid hunters in re- triexing wounded geese that had been knocked down in the fields. Through this device, hunters secured a fair number of birds that would otherwise not have been recovered. Crippled geese within the refuge were usually found apart from the fl\ing birds, sometimes gathering in flocks of 10 or more. The strongest cripples swam about in the lake, where they sought shelter close to shore among the cypresses and snags, fig. 42, but the weaker ones rested on the lake shore. Few badly shot geese re- covered from their wounds ; many sur- vived for a time, but in their weakened condition they became victims of predators. Raccoons consumed many dead geese (Yeager & Elder 1945). Although un- able to catch healthy birds, these animals apparently sought out and killed many of the cripples, the remains of which were usually found along the shore line or on logs some distance from shore. A few carcasses were pulled under water and eaten by turtles. Skeletons of many geese have been observed on the lake bottom in years when cripple surveys have been made on ice. Undetermined numbers of geese sought shelter and died in parts of the lake that were inaccessible to man be- cause of the large number of dead trees and fallen logs. Some cripples were caught by foxes and dragged into the woods on the refuge, where they were devoured ; others died on hunting lands away from the lake. Consequently, a count of skele- tons and carcasses around the shore line and on the island and club grounds repre- sented only a portion of the total loss. To determine at least the minimum number of unretrieved geese that died of wounds, counts were made of goose car- casses along both island and outer shore lines of the lake, as well as on the grounds of the principal goose clubs. The total counts of carcasses each winter, from 1940—^1 through 1945-46, are given in table 20. Not all carcasses counted, of course, represented cripples that had died, IS8 Illinois Natural History Survey Bulletin Vol. 25, Art. .1. Tiible 20.—Number of carcasses of Canada geese counted on and near the Horseshoe Lake Game Refuge, 1940-41 through 1945-46. .March. 1950 H.ANSox & Smith: Canada Geese 159 advanced increased the likelihood that geese would pick up shot in fields planted to winter wheat, one of their principal foods in the Horseshoe Lake area. In areas where the ground held much mois- ture and the wheat plant was devoured down to the roots, some soil and probably any shot that happened to be present were ingested. Geese at the refuge were ob- served to devour considerable quantities of soil at times, particularly in winter. In certain farmed feeding areas, holes as much as 6 or more inches deep and several times as wide were created by the geese in their ostensible search for food. This type of feeding increased the likelihood of the birds occasionally swallowing lead shot. "Tip-up" feeding by Canada geese in the water of the Horseshoe Lake area was observed in late winter. This habit may have been a response to a reduced food supply on land. In 1942. a slough on the west side of the lake was a favored "tip- up" ground. Dr. William H. Elder, while with the Illinois Natural History Sur- vey, when surveying this area for cripples, found 13 dead or dying geese on the ice or close to the shore line. Of 23 geese autopsied by Dr. Elder in late winter, 20 were found to have died of lead poisoning, 18 of these containing shot in their gizzards. Paul S. Smith of the United States Fish and Wildlife Service, who conducted a series of tests on one of the most heavily shot club grounds, found about one lead shot per square foot of top soil, 1 inch in depth. Only the fact that the grounds of hunting clubs were cultivated each year prevented losses due to lead poisoning from assuming greater proportions. The po- tential danger from lead shot increased each year of the study, and the proximity of heavily shot fields to such an important concentration area as the Horseshoe Lake Game Refuge constituted a significant hazard to the geese wintering there. Starvation A Canadian Indian whose hunting grounds lie in the Lawapiskau River* country related that during late springs, when snow remained on the ground for some time after the arrival of Canada • This river flows into James Bay 20 miles soulh of die Albany River. geese, he found dead birds that were in a very emaciated state, a condition that he attributed to a lack of available food. Be- cause it is likely that, as a result of disease, lead poisoning, or crippling, a few geese succumb soon after their arrival on the breeding grounds, it is impossible to assess from this single report the importance of starvation as a cause of death in Canada geese. Nevertheless, there is some evidence that a food shortage in late spring may re- sult in death of the weakest birds. In the second week of May, 1947, when the rivers and creeks were frozen and tlie country was still under se\eral feet of snow, geese shot by Indians at the south end of James Bay were reported as having only willow catkins in their gizzards. Bound Crop Occasionally Canada geese were found in the vicinity of Horseshoe Lake in a thin, weakened state and with greatly Fig. 65.—Esophagus, proventriculus, and gizzard of a Canada goose found dead on the Bright Land Farm near Barrington, Illinois. Death in this case was due to lead poisoning from 38 shot found in gizzard. Food impac- tion is the result of lead poisoning, which often causes paralysis of the digestive tract in Canada geese and other waterfowl. (Pho- tograph by Charles W. Kossack.) distended crops. Examination of these in- di\ iduals re\ealed that an impacted crop was often the primary cause of their con- dition, and, though operative measures were tried, few of these geese had suffi- cient stamina left to survive. Their crop contents usually consisted of a tightly packed mixture of wheat browse, corn, 160 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 and cou-peas, or sojbeans, and frequently leaves and portions of the stems of the two legumes. In some of these, bound crop may not have been the direct cause of loss of weight and strength; instead it may have been the result of partial paral- ysis and weakness resulting from lead poisoning, fig. 65. C. E. Laughery, formerly refuge mana- ger at the Horseshoe Lake Game Refuge, informed us that geese with bound crops were most frequently found in years when several weeks after most of the local corn crop had been utilized or removed from the fields, fig. 66. While consumption by geese of shattered and otherwise wasted soybeans may seem desirable, these beans may sometimes have contributed to a num- ber of deaths resulting from bound crop. In the winter of 1943—44 in particular, Paul S. Smith, when surveying the vicinity of the refuge, found a number of dead geese, their crops tightly packed with soy- beans. These birds were said to differ Fig. 66.—Canada geese in a harvested soybean field near Horseshoe Lake, autumn 1946. a considerable acreage on the refuge was planted to cowpeas. Cowpea fields attract large numbers of geese long after the bulk of the crop has been consumed. A few geese, while searching for peas, evi- dently consume fibrous and relatively in- digestible portions of the plant. The pres- ence of such material in the crops of geese may be responsible for impactions. In recent years, soybeans were planted extensively in southern Illinois, and the geese tended to utilize this crop to a greater extent each year. There was fre- quently much wastage in harvesting these beans; many fields in the vicinity of Horse- shoe Lake were not combined until an appreciable portion of the crop had been lost through shattering. As a result, beans in abundance were available to geese for from the crop-bound birds described above in that they were particularly heavy and fat. Probably in the winter certain geese fed more extensively on soybeans than on other foods and, as soybeans have a high protein and fat content, these individuals became heavier than the average goose of the area. Apparently these geese died after drinking water when their crops were crammed with beans. The pressure re- sulting when the beans imbibed water and swelled may have been the direct cause of death in such cases ; the mechanism of the lethal effect is not known to us. Geese frequently stuff their crops tight with corn, but in only one instance was corn suspected of being an indirect cause of death. This individual with an over- loaded crop, fig. 67, became agitated in March, 1950 Hanson & Smith: Canada Geese 161 Fig. 67.—The Canada goose is a voracious eater. This iiulividual with an overloaded crop became frightened when a game techni- cian entered the trap in which it had been caught. It hail extreme difficulty in breathing and died a few minutes later. the trap and died shortly thereafter, e.x- hibiting the syndrome typical of ano.xia. Rough tests of the swelling properties of dry soybeans and corn revealed that the beans present a much greater hazard as food for geese than does the corn. Soy- beans and corn were soaked in water for intervals varying from 30 minutes to 6 hours. Water displacement measure- ments showed that the soybeans increased their bulk at a rate appro.ximately three times the rate corn increased its bulk. At the end of 3 hours, soybeans had in- creased their bulk by 85 per cent and corn by 30 per cent. These data and field ob- servations suggest that soybeans and cow- peas may not be ideal crops to plant for the e.xpress purpose of providing food for wintering concentrations of Canada geese. Predators The red fox is probably the only preda- tor at Horseshoe Lake that is capable of catching sound, healthy geese. Remains of geese found in cornfields late in the autumn point to predation by foxes, but probably most carcasses represented secondary predation involving birds crip- pled during the hunting season. In each year covered by this study, a pair of bald eagles nested on the island in Horseshoe Lake, and both adults and juveniles were observed regularl\- through- out the autumn and winter periods. In the autumn of 1945, the eagle population on the refuge numbered at least five. Eagles were frequently seen feeding on crippled geese that had died, and in December, 1945, several eagles were ob- served by Paul S. Smith to attack a live goose (probably a weak cripple) that was frozen to the ice by its feet and breast feathers. Eagles were ne\er seen to at- tack a sound, healthy goose. Bald eagles are reported to feed on wounded geese in the Port Joli area of Nova Scotia, and never to be absent from the area as long as the geese remain (Tufts 1932). A discussion of predators on the breeding grounds will be found in the section on "Productivity." Diseases Only two diseases were investigated at Horseshoe Lake: tracheitis and asper- gillosis. Tracheitis.—In January, 1945, a number of geese trapped were found to have wheezy voices, indicative of a con- gested tracheal condition. Two of thete birds eventually died, and the lungs and trachea of one were sent to the Depart- ment of Animal Pathology and Hygiene, University of Illinois, for examination. The cause of death was diagnosed as tracheitis, pulmonary congestion, and edema. The symptoms of the disease as obser\cd at Horseshoe Lake were a \oice pitched higher than normal, a distinct "wheeze," and heavy, spasmodic breathing, accom- panied by a forward throw of the head and open mandibles as the bird gasped for air, fig. 68. As the disease progressed, the effort attendant upon the intake of Fig. 68.—Canada goose near death frotn tracheitis. Symptoms of this disease are a forward throw of the head and neck and gaping as the bird gasps for air. 162 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 air became increasingly spasmodic and violent because of the whitish exudate that accumulated in the trachea at the junction of the bronchi. Both field experience and laboratory findings indicate that tracheitis is infec- tious, but the nature of the infectious agent is uncertain. Graham & Thorp (1931) have reported that a Canada the typical nodules associated with Asper- gillus infections were present throughout the body cavity, fig. 69, left. In December 1946, a second juvenile goose was found dead from an Jspergillus infection. Post- mortem examination of this specimen by the Department of Animal Pathology and Hygiene revealed that the air sacs were partly, or in some cases completely, filled Fig. 69.—Aspergillosis in Canada geese. The nodules of Aspergillus infection shown in the illustration at left are on the lateral wall of the body cavity. In the goose shown in the illustra- tion at right, the air sacs are the principal foci of infection. Both specimens were juveniles. goose from a farm flock had clinical symptoms analogous to acute laryngotra- cheitis in domestic fowl. However, au- topsy of the goose revealed that the lung contained foci of mycotic pneumonia. Aspergillosis.—The manifestations of aspergillosis in waterfowl have been ade- quately described by Phillips & Lincoln (1930). While outbreaks are known to occur occasionally in duck populations (Phillips & Lincoln 1930; Pirnie 1935; Bellrose, Hanson, & Beamer 1945), only one instance of its occurrence in Canada geese in the wild has been recorded pre- viously (Dow 1943). At Horseshoe Lake on November 7, 1946, a juvenile Canada goose was found in a much weakened condition. Within a day it was dead, and autopsy revealed that with a fungus growth, fig. 69, right, that upon cultural examination presented the cliaracteristics of Aspergillus fumigatus. Parasites Both interna! and external parasites were taken from Canada geese wintering at Horseshoe Lake. External Parasites.—Four species belonging to four different genera of chewing lice or Mallophaga were taken from Canada geese at Horseshoe Lake. Specimens of Trinoton querqueduhte Linnaeus collected in the winter of 1945— 46 were identified by Dr. Carl O. Mohr, then of the Illinois Natural History Sur- vey staff. The following species, collected from a dead goose in 1934, were identified by R. O. N. Malcomson: Anatoecus March, 1950 Haxson & Smith: Canada Geese Ibi ferruyiiieus Giebel, Esthiopterum crassi- corrte (Scopoli), and Ornithobius gonio- pleurus Denny. Internal Parasites.—Flukes were frequently encountered in the cloacae of Canada geese at Horseshoe Lake when examinations were made for sex and age. A number collected in the winter of 1945— 46 were referred to Dr. L. J. Thomas of the Department of Zoology, Univer- sity of Illinois, for identification. In his report he identified these specimens as Echinosloma revolutum and Prostliot/oiii- mus sp. ; specific identification in the lat- ter genus was impossible because of the poor condition of the specimen. PRODUCTIVITY' It is important to know several months in advance the probable population of any game species at the start of a hunting sea- son in order to determine what hunting restrictions will be necessary in that sea- son. Populations of nonmigratory game can be estimated or inventoried before the hunting season more easily than can those of such migratory species as the Canada goose, which nests in compara- tively inaccessible regions. Because of the length of time generally required be- fore they can be officially approved, hunt- ing regulations for migratory waterfowl must be decided upon while the actual size of the fall population is still an un- known. Thus, it is desirable to be able to forecast the population accurately from data obtained during the previous season. Forecasts can be made more easily for a population of limited size and distribution, such as the Horseshoe Lake goose flock, than for immense, continent-wide popu- lations. To interpret and predict population trends from flocks on their wintering areas, such questions as these must be answered : What is the age ratio, within the flock, of juveniles to adults? What are the survival rates of various age and sex groups? How long do geese live? How many or what percentage of a popu- lation attain breeding age? What is the ratio of males to females? Does a dis- proportionate kill occur in the various sex and age groups ? Answers to these ques- tions have been sought in studies of the Canada goose at Horseshoe Lake and on the breeding grounds, and in records of geese banded at the Miner Sanctuary. Breeding Potential The theoretical capacity of a species to produce young is determined by mating habits, age at reproductive maturity, ratio of males to females, and number of young produced per season. Information in the literature on these subjects is briefly sum- marized to aid in interpreting the signifi- cance of related data from the Mis- sissippi fiyway. Mating Habits.—The Canada goose is monogamous and, judged from the habits of captives, fig. 70, remains paired to the same mate as long as both are alive. In captivity, individuals have been known to re-pair after the death of a mate (Montgomery 1938), although in some cases several years may elapse before re- mating takes place (Miner 1923). Re- mating experiments with Canada geese by Charles Kossack and Carleton Beckhart at tlie Bright Land Farm near Barring- ton, Illinois, have shown that a very high percentage of captives will remate the first spring following separation from their mates. Reproductive Maturity.—At least 2 \ears are required for the Canada goose to reach sexual maturity in the wild, and in captivity the age of maturity is often 3 years and sometimes 4 (Dutcher 1885, Bailey 1913, Taverner 1922, Wilfrid 1924, and Forbush 1925). Studies made by the Illinois Natural History Survey of the semicaptive flock at the Bright Land Farm revealed that 25 per cent of the geese bred during their third year (Elder 1946). Definite information on Canada geese breeding in the wild at 2 years of age is lacking. If the presence of an open oviduct is a sign of sexual maturity or an indica- tion that eggs have been produced, data from Horseshoe Lake indicate that in the wild practically all females are productive at 2 years of age. However, until further information is available, inclusion of all wild geese in their third year of life in the breeding component of the population must be considered tentative. Of 54 females banded as juveniles and retrapped and examined at Horseshoe Lake in their 164 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Fig. 70.—Female Canada goose and newly hatched young on the Bright Land Farm near Harrington, Illinois. (Photograph by Charles W. Kossack.) second winter, all possessed closed oviducts (at about 11,-4 years of age); but of 18 females retrapped and examined in their third winter (at about 2l^ years old), all but one possessed open oviducts (Hanson 1949fl). The duration of fertility is probably not a factor limiting the productivity of Canada goose populations, as captives have been known to raise young at ages that far e.xceed the length of life of most individuals in the wild, few of which live longer than about 5 years (see section on "Population Survival"). Sex Ratios.—Sex ratios of Canada geese as they were obtained from trapping and from bag inspection in the vicinity of Horseshoe Lake are given in tables 21 and 22. In the juvenile age class, trap data for the period of study indicate a slight but statistically significant excess of males ; bag data, on the other hand, indicate no March, 1950 Hanson & Smith: C.an.-\d.\ Geese 165 Table 21.—Number of male and female juvenile Canada jieese newly trapped and banded and number examined in bag at or near Horseshoe Lake, 1940-41 through 1946-47. 166 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 The number of eggs produced by cap- tives is surprisingly close to the production attained by wild birds. In 1Q42, 54 pairs of Canada geese on the Bright Land Farm produced 250 eggs, or an average of 4.81 eggs per pair. Several people experienced in raising Canada geese have stated that the number of eggs laid may vary with the age of the birds. Dutcher (1885) cites a game breeder on Long Island who claimed that 4 eggs are laid the first year of breeding, 5 the second, and 6 or 7 thereafter. Miner (1923) also states that "a j'oung goose will lay four eggs the first year [of laying] and usually five the second." Actual Productivity The number of young birds brought to the flying stage is always somewhat less than the theoretical maximum. Fertility of Canada goose eggs is evidently high. In California and Utah, an egg fertility of 93 and 94 per cent, respectively, was found. However, flooding, predators, and other agents may destroy as high as 40 to 48 per cent of the nests in Cali- fornia (Dow 1943) and thus reduce pro- duction of young. Consequently, the an- nual production for all pairs that nest may average only 2.48 to 2.84 goslings per pair, or about 50 per cent of the number of eggs produced. In Utah, 84 nests studied yielded an average of 3.9 goslings per nest (Williams & Marshall 1938). Second nestings are sometimes attempted, a factor that would somewhat increase the average annual productivity per pair. Information volunteered by the Indians at Moose Factory, Fort Albany, and At- tawapiskat suggests that the red fox is the predator most destructive to Canada goose nests in the James Bay area. The extent to which foxes are harmful to goose nests is probably inversely related to the population levels of other prey species. In 1946, a year during w-hich foxes were abundant, but snowshoe hares, muskrats, grouse, and ptarmigan were low in num- bers, Indians reported finding many Canada goose nests destroyed by foxes. When interviewed in the summer of 1947, one Indian said, "The foxes are now low in numbers. Let's wait and see what kind of luck the geese have in raising young this year." These attitudes by a native people, who are the keenest of observers, should be given careful consideration. Recent studies have generally confirmed the belief that predators have little effect in controlling the numbers of cyclic prey species, but in the case of Canada geese we are dealing with a bird that is normally of secondary importance as a prey species and that at present is not known to be cyclic. If geese and other waterfowl are subject to increased predation by foxes when these animals are at the peak of their cycle, it is conceivable that the numbers of water- fowl could be measurably affected by fox predation. A few Indians that remain in the in- terior occasionally take goose eggs, but as the greater number of the Indians are at the coastal posts, fig. 55, during the nest- ing season, the importance of Indian pre- dation is negligible. Juvenile mortality in Canada geese ap- pears to be small. In Utah a 3 per cent decrease in average brood size occurs over a period of a month (Williams & Marshall 1938). Little is known concerning pre- dation on broods, but in one recorded instance in British Columbia ring-billed gulls devoured a brood of newly hatched goslings (Munro 1936). The scarcity of natural enemies in the James Bay muskeg area normally insures small losses of goslings to predators ; coyotes are absent, wolves almost non- existent ; lynxes, minks, martens, fishers, and otters are generally scarce, and wol- verines are extremely rare. Probably foxes, abundant at the peak of their cycles, are predators of consequence only in years in which populations of snowshoe hares and other prey species are low. Great horned owls are fairly common and may account for the loss of a few young geese. Data From Horseshoe Lake The degree to which goose productivity measurements at Horseshoe Lake are a valid measure of the actual productivity of the Horseshoe Lake flock on the breed- ing grounds is dependent upon the magni- tude of the losses between the James Bay area and Horseshoe Lake (see section on "Annual Bag"). The autumn kill by the Canadian In- dians is small, fig. 61, so that, even if more .March, 1950 Hanson & Smith: Canaoa Geese 167 Fi^. 71.— lype of trap used to catch Canada gtese at Horseshoe Lake. Trap consists of eight wood and wire frames roofed over with twine netting and supported with guy wires. Open ends are closed off by tripping a pipe-weighted, twine curtain from a blind. young than adults are killed in proportion to their numbers, the ratio of juveniles to adults in the flocks is not changed ap- preciably by the time the geese migrate southward. The scarcity of band re- coveries between James Bay and southern Canada further indicates that the flocks are still largely intact when they reach the northern border of the United States. From fig. 38 it is evident that the majority of the geese have arrived at the Horse- shoe Lake Refuge by November 1. As the bulk of the kills north of the refuge are made after this date, fig. 62, most of the flocks that arrive at the refuge have been only moderately depleted b\' shooting ; band recoveries indicate that total hunting losses between the Canadian border and Horseshoe Lake are usually 5 to 6 per cent of the southward bound population. Because of the small migration losses in the population, the over-all ratio of young Mi-Mi i IS- 72. — Canada geese feeding into drop curtain trap at Horseshoe Lake. 168 Illinois Natur..\l History Survey Bulletin Vol. 25, Art. 3 to old in the flocks as they arrive in southern Illinois probably does not differ greatly from ratios existing at the time the family groups start their southward migrations; therefore, we believe that age and se.x ratios at Horseshoe Lake furnish reasonably accurate measures of actual productivity ratios in most years. Trapping data and data from bag in- spection and band recoveries combined have been used in measuring production of the Canada geese wintering at Horseshoe Lake. The age ratios of the geese caught in traps are believed to be fairly representa- tive of the untrappd population for the following reason: No significant dif- ference was observed in the wariness of geese of the various age classes as the birds entered the traps (many geese were color-banded to indicate age classes). Many catches consisted of individuals that had entered the traps, fig. 71, as parts of a busily feeding wedge ; such catches would not represent selective trapping, fig. 72. The ratio of juveniles to adults during the early part of the autumn no doubt differs to some extent from the ratio after the hunting season because of the propor- tionately greater kill of juveniles, table 23, but we are not able to demonstrate the extent of this difference from the data at hand. The total annual catch since the win- ter of 1943-44, excluding repeats, includes a large percentage (5.4, 22.5, 42.7, and 30.0 per cent, table 1 ) of geese trapped and banded in previous years (trap re- turns), many of which are accompanied by their unhanded young. Therefore, the ratio of juveniles to adults among the newly banded birds, table 23, is not in- dicative of the age ratios in the flock as a whole, as it necessarily excludes the many banded adults that returned to the traps. The ratio of juveniles to adults for entire-season catches is given in table 24. The figures for the season catches in this instance include the geese trapped in a specified season but banded in a previous season (trap returns) as well as the newly banded birds, but they exclude birds banded and retrapped in the same season (repeats). These data more nearly rep- resent the actual juvenile-adult ratio in the flock than do the data on newly banded geese given in table 23. Data from bag inspection are indicative of true flock ratios only when they are corrected for differential hunting vulner- ability of the juveniles by means of trap and band-recovery data. Age ratios de- rived directly from band-recovery data do not accurately reflect the age ratios in the total population for the same reason that the age ratios of unhanded geese in the hunters' bag do not, namely, that the banded juveniles are shot more heavily in proportion to their actual numbers than are the banded adults. However, age ratios derived from band recoveries can be used to correct bag ratios for the dis- proportionate kill of juveniles as follows : (1) Determine the relative vulnerability to shooting of the juveniles and the adults. (2) Use the vulnerability quotient of the Table 23.—Number of juvenile and adult Canada geese newly trapped and banded and number examined in bag at or near Horseshoe Lake, 1940-41 through 1946-47. March, 1950 Hanson & Smith: Canada Geese 169 Table 24.—Productivity of the Horseshoe Lake flock as shown by trapping and bag- inspection ratios.' Season 1940-41 . . 1941-42. . 1942-43. . 1943-44 1944-45 . 1945-46. . 1946-47. . Tolal . . .Average From Trapping Total Num- ber of Indi- viduals Trapped of Known .Age-Cla.ss and Sex'' 313 408 1,054 2,462 1,101 541 717 6,596 Number of Breeding Adult Females Trapped^ 136 88 114 33S Number of Juveniles Trapped 143 274 619 1,379 607 196 296 J, 514 Juveniles "per 100 Adult Females 446 223 260 0^9 325 Juveniles per 100 Yearlings and Adults 84 204 142 127 123 57 70 807 U4 Fro.m Bag Inspection ' Sec page 168 for explanation of reasons figures in tiiis table difTer from those in table 23. ' Numbers in this column include returns (geese banded in previous years). ' About 2y!t or more years old at time of trapping. Juveniles per 100 Yearlings and Adults 126 juveniles to correct bag ratios for the disproportionate numbers of juveniles lost through shooting. The vulnerability quotient of the juve- niles is obtained b\ the following formula, suggested by Frank C. Bellrose: Number of band re- coveries from juve- niles Vulnerability quotient V = Number of juveniles banded before end of hunting .season Number of band re- coveries from adults Number of adults banded before end ot hunting season According to these calculations, at Horseshoe Lake in 1943, the juveniles were 8.34 times as vulnerable to shooting as were the adults. With this figure avail- able, it is possible, assuming the vulner- ability quotient to be a true measure of vulnerability of the juveniles, to correct the age ratios obtained from bag inspec- tion, which, by \'irtue of the higher vulner- ability of the juveniles, is weighted in favor of this group as coinpared with the adult group in the total surviving popu- lation. To correct age-ratio data obtained from bag inspection, it is assumed that the fol- lowing formula is true, in which V is the vulnerabilit.\' quotient calculated above from the trap and band-recovery data. Data that are perhaps numerous enough to use in determining the vulnerability of the ju\ eniles as compared with the vulner- ability of the adults are available only for the 1943 hunting season. In 1943: Ratio Number of juveniles in bag Juveniles in population Number of adults in bag .Adults in population 170 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Ratio (1943) Ratio (1943) 158 Juveniles 15 Adults 8.34 Juveniles 10.53 = 1.26 Adults 8.34 Then the corrected age ratio is 1.26 juveniles to 1.0 adult. The ratio of juveniles to adults found above from corrected bag ratios for 1943 is close to the age ratio found from trap- ping for 1943 (127 juveniles to 100 adults, table 24). Over a 7-year period the juvenile age class comprised about 53 per cent of the birds in the Horseshoe Lake flock, table 24. Not only is it important to know what percentage of the flock is composed of juveniles each year for a significant anal- ysis of productivity; it is important to know also the production of young in re- lation to the number of mature females — birds that are 2i/2 or more years old when wintering at Horseshoe Lake. By re- lating productivity to only the sexually mature females, compensation can be made in statistical analyses of the flock for annual changes in the percentage of non- breeding yearlings as well as for changes in ratio of adult males to adult females. These productivity figures will be at variance with the impression that the aver- age hunter gets from the flock at Horse- shoe Lake. This hunter, on viewing the impressive concentration of geese at Horse- shoe Lake, thinks that the total number of mated pairs in the flock in the following spring will equal the total population divided by two. Since he has heard that geese annually lay 5 or 6 eggs, he assumes that there will be an impressive increase for the next hunting season, and, thinking in terms of himself, anticipates more shoot- ing. When informed that the flock may be even smaller in numbers at its peak in the autumn than it was at the close of shooting the previous year (as actually happened in the autumns of 1944 and 1945), in spite of the young added to the flock as a result of the breeding season, he may be dubious as to the competence of his informer. The layman often fails to take into ac- count the fact that Canada geese do not Table 25.—Age and sex composition of the Horseshoe Lake flock, 1944-45 through 1946—47, as shown by trap catches of unhanded and previously banded geese. March, 1950 Hanson & Smith: Canada Geese 171 breed until they are at least 2 or 3 years old, and that at least one-half of the birds he sees will still be sexually immature in the spring ; that an excess of males exists in the birds of breeding age, table 25 ; that members of broken pairs may be slow to mate; that some pairs each year are not successful in rearing a family ; and that natural losses as well as the Indian kill are taking place in the intervening months. The actual number of sexually mature fe- males upon which production in the coming spring is dependent may comprise only a small segment of the winter flock. in some years as low as 12 to 17 per cent, table 25.' A rapid method of distinguishing year- lings from older geese, for use on live birds in the field, was not developed until the fall of 1944 (Hanson 194%). Since the more nearly complete data from 194-1— 45 through 1945—1-6 were collected dur- ing and after hunting seasons in which higher rates of loss occurred among juveniles than among adults, the actual ratios of juveniles to breeding females existing before the shooting began would be somewhat higher than those indicated in table 24 ; the trap ratios in table 24 differ from the indicated ratios or percent- 4- 172 Illinois Natural History Survey Bulletin Vol. 25, Art. 3I pressed as a percentage of the total ob- servations. From these data we suggest : 1. That the average family-flock size in late summer or early autumn may fur- nish a rough index of the age ratio within a large population, fig. 73 ; frorn this ratio the success of nesting the previous spring mav be inferred, A and D in fig. 64. 2. That the average family-flock size in middle or late autumn, when compared with similar data gathered the same year before the opening of the hunting season, is indicative of the degree to which family groups have been broken through shoot- ing. (Compare F and G in fig. 64.) POPULATION SURVIVAL* One of the objectives of the Canada goose research program reported in this paper was to determine through trapping and banding the annual mortality rate and the average longevity of Canada geese in the Horseshoe Lake flock in the period 1941—1946 and to compare the annual mortality data derived from the banding at Horseshoe Lake with similar data de- rived from the banding of Mississippi fly- way geese at the Jack Miner Sanctuary in the period 1925-1944. Definition of Terms In the following discussion, age class refers to a group of geese, all of them hatched in a given year. A banding class includes all geese banded in a given season regardless of age at time of banding. The computed percentage of geese of a banding class alive each year in a series of succes- sive years following banding comprises a survival series. This series may be com- puted from data in a band-recovery series (recoveries of bands from birds reported dead in any of several successive years after banding) or from data in a trap series (returns of banded birds to the traps in any of several successive years). • The senior aullior is responsible for this section. As he carried out the trapping program at Horseshoe Lake and the compilation of the Miner recovery data in Ottawa, Canada, he is fully conscious of the inade- quacies and bias in the data on which the following discussion is based. Tiiese inadequacies and bias do not permit the data to he treated by the customary metliods. Tlie methods used by-pass some of the shortcomings of the data, but. in the final analysis, the results presented only produce an approximation of the true picture. The reader should bear this point in mind in evaluating the results presented. It was deemed advisable to exploit the data as far as possible rather than disregard them altogether because of an acute awareness of their vagaries. Table 26.—Hypothetical catches of Canada geese to illustrate difference between a trap series and a return series. March, 1950 Hanson & Smith: Canada Geese 173 Table 27.—Approximate mortality of juvenile Canada geese durin;^ the first year after banding (vear 0-1) at Horseshoe Lake, as determined bv censuses and age ratios from trapping, 1943-44 through 1946-47. 174 Illinois N.atur.al History Survey Bulletin Vol. 25, Art. 3 results in catches that are composed of disproportionately large numbers of band- ed yearlings in relation to the actual num- bers of this age class in the untrapped por- tion of the population, with the result that calculated mortalities for yearlings, al- though seemingly very high, may be below the rates that actually occur in that age group in the unhanded segment of the population. In table 28 the actual number of trap returns from geese banded as juveniles at Horseshoe Lake is given, and in table 29 these numbers have been converted into percentages of the original bandings. For example, 67 geese banded as juveniles in the trapping season of 1942—43 and 13 geese banded as juveniles in the season of 1941—42 were trapped in the winter of 1943—14, table 28. Expressed as percent- ages these returns were 10.82 and 4.74 per cent of the original bandings (619 and 274, respectively), table 29. The survival series figures in tables 29 and 32 were derived from the weighted average per cent returns of geese of con- secutive }'ear classes, beginning with year 1-2 (first trapping season after year of banding). Tables 28, 29, and 30 include only juvenile-banded geese; tables 31, 32, and 33 include both juvenile-banded and adult-banded birds. A more nearly accurate picture of sur- vival in age classes than that given by table 29 begins with the 26.46 year-of-banding survival figure, table 27. Survival in subsequent years was derived from the weighted average per cent returns in table 29 through the formula explained in foot- note 4 of that table ; the survival series figure for the year previous to year 1-2 is assumed to be 26.46. For example, 9.57:4.17::26.46:x; x is 11.53, the sur- vival series figure for the year 1-2. The entire survival series is 26.46, 11.53, 3.90, 3.73, 1.99, 3.87, fig. 74. The weighted average survival rates, as calculated from this survival series by the method suggested Table 29.—Trap returns of Canada geese banded as juveniles at Horseshoe Lake, ex- pressed as percentages of original bandings, 1940-41 through 1946-47. Season March. 1950 Hanson & Smith: C.an.vo.x Geese 175 Table 30.—Annual mortality rates (per cent) of juveniles in the Canada goose flock at Horseshoe Lake. 194l)-41 through 1946-47. (See formula, page 173, and data in table 29, top). Season 176 lOOH Illinois Natural History Survey Bulletin Vol. 25, Art. 3 80- tiJ > 60H LiJ O 40 (T LiJ Q. 20- I BANDED IN FIRST YEAR OF LIFE n BANDED IN FIRST OR LATER YEAR OF LIFE -1— 3 —r- 4 -I— 5 -r- 6 T 7 YEAR AFTER BANDING Fig. 74.—Survival of two groups of Canada geese, one banded in first year of life and one banded in first or later year of life. Curve I figures are from page 174; Curve II figures are from table 32, survival series 1. geese remaining alive in each successive year. The survival rates cited above in- dicate that an average of 74 per cent of the original bandings disappeared by the end of the year of banding, 56 per cent of the survivors were lost during the second year after banding, 66 per cent the third year, 4 per cent the fourth year, and 47 per cent the fifth year, table 30. Data on returns from banded geese of all ages, that is, the combined returns of birds banded as juveniles, yearlings, and geese of unknown age, have been treated in the manner described above, tables 31, 32, and 33. The survival series obtained, 49-22-11-7-5, table 32 and fig. 74, is believed to represent the approximate rate at which the average banding class in the Horseshoe Lake flock disappeared during the first 5 years of life following banding in the trapping seasons 1940-41 through 1946-47. The disproportionate loss of juveniles that usually occurs, in large part from shooting, does not weight this por- tion of the survival series, since the series is based on the total per cent of the geese returning to the traps 1 or more years after banding. The weighted average per cent return, 9.70, table 32, of geese the first year after the year of banding neces- sarily represents birds that are at least Ij/z years of age. The calculation methods discussed above leave much to be desired, particularly those involving mortality rates of the juveniles during the first year of life after March, 1950 Hanson & Smith: Canada Geese 177 Table 33.—Annual mortality rates (per cent) in the Canada ijoose flock wintering at Horseshoe Lake, 1940-41 through 1946-47. (See formula, page 173, and data in table 32, top.) Season 178 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Table 35.—Recoveries in the Mississippi River valley of bands from Canada geese banded at Kingsville, Ontario, in the autumn, 1925-1932. The recoveries are for 12 years, includinfi the year of banding.' 1925-1932 March, 1950 Hanson & Smith : Can.ad.a Geese 179 ing, as well as during subsequent banding seasons, and (2) the tendency of some in- dividuals to establish a trap habit that persisted in later years. For several reasons it seemed desirable to make an "across the board" treatment of the trap data, that is, an analysis of mortality from annual random samplings of the retrapped banded survivors. Tables 28-33, referring to trap returns, should be read horizontally ; they should not be read diagonally, as they would be if a single banding class were followed through the years. A few geese banded at Horseshoe Lake winter in parts of the Mississippi flyway other than at this lake, and while some disperse to other flyways, table 4, there is no evidence that this dispersal to a dif- ferent wintering range is greater during any particular year than in others, a factor that might otherwise seriously influence the validity of our sur\i\ al series. Mortality Calculated From Band Recoveries.—The survival rate meas- ured by the use of band recoveries is based on the assumption that the unhanded segment of a population disappears at approximately the same rate as the banded segment and that year-to-year differences in the numbers of banded birds reported dead in successive years is indicative of the annual mortality of the entire popula- tion. However, unless all banding is Table 38.—Recoveries in the Mississippi River valley of bands from Canada geese of all age classes banded at Horseshoe Lake, 1940-41 through 1944-45. Trapping 180 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 coveries reported during each succeeding year after banding is subtracted from the number of banded geese unrecovered and presumably alive the preceding year; then the number of geese unrecovered and as- sumed to be alive in each year is ex- pressed as a per cent of the total recoveries, table 35. The second method must be used for recovery data from bandings at Kingsville, Ontario, because the size of the original banding is not known with certainty and because an unknown portion of the bandings listed in table 2 were Mississippi Valley geese ; presumably the remainder represented the Southeast popu- lation. For the purpose of comparing mortality in another segment of the Mississippi Valley population since 1925 with mortal- ity in the Horseshoe Lake Hock, recoveries of geese banded at the Jack Miner Bird Sanctuary in the autumn were used, table 34. Although band recoveries from geese of unknown age at the time of banding do not give a precise picture of population mortality in Canada geese because of the differentially high kill of the juveniles by hunters, they suffice as a basis for a com- YEAR OF BANDING INCLUDED YEAR OF BANDING EXCLUDED -r 4 T 73 4 5 6 7 8 YEAR AFTER BANDING 9 10 Fig. 75.—Average survival of Mississippi Valley Canada geese, as measured by band re- coveries from geese banded at the Jack Miner Bird Sanctuary, Kingsville, Ontario, in the autumn, 1925-1932. Curve I includes band recoveries made during the year of banding; curve II excludes recoveries made during the year of banding. Curve I (data from table 35) starts with an expression (100 per cent) of the total number of recovered bands; curve II (data from table 41) starts with an expression (100 per cent) of the total number of recovered bands that were on geese alive at the beginning of the year following banding. I March, 1950 Hanson & Smith: Canada Geese ISl 925-29 1930-34 935-39 YEAR AFTER BANDING Fig. 76.—Average survival of Mississippi Valley Canada geese, as measured by band re- coveries from geese banded at the Jack Miner Bird Sanctuary, in the autumn, 1925-1939. That part of each curve representing the year of banding shows a higher rate of survival than actually occurred, as in the data (from table 36), which represent the number of bands re- covered and not the number of bands applied; no correction was made for the varying lengths of exposure to guns experienced by geese banded at various times in the season of banding. parison of mortality rates in different years. As no individuals from the Miner autumn bandings have been reported shot in the Mississippi River valley later than 12 years after banding, recoveries of geese banded in 1925—1932 may be considered nearly 100 per cent complete by 1944. These data, summarized in table 35 and presented graphically in fig. 75, curve I, show that maximum sur\ival in Canada geese in the Mississippi River valley under moderate hunting pressure is about 12 years. Since about 93 per cent of the bands in the 12-vear series were recovered bv the end of the si.xth or seventh years after banding, table 35, no great error would result from basing an analysis of mortal- ity from 1925 through 1939 on the num- ber of banded geese reported dead by the end of the sixth or seventh years. Re- coveries of birds banded in those years are grouped by three 5-year periods. These 5-year data groups are set off by horizontal lines in table 34. In table 36, they have been summarized. The survival curves based on these data are shown graphically in fig. 76. In order to compare the survival of geese banded at the Miner Sanctuary in 182 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 1940-1944 with the survival achieved by geese banded by the Miners in previous years, it was necessarj- to use an incomplete band-recovery series, derived from table 34, as explained in a footnote to table 37. The groupings for this analysis are sum- marized in table 37 and the computed survival curves are shown in fig. 77. First-year survival data obtained from recoveries of geese banded at the Miner Sanctuary in the autumn are not an ac- curate representation of average first-year survival for Mississippi flyway geese as a whole. Whereas most bandings of water- fowl yield the greatest number of re- coveries during the year of banding, the largest number of recoveries from Miner bandings have been received in most in- stances the year following the year of banding, table 34. One reason for this situation may be that the geese that are trapped and banded represent those that remain at the sanctuary the longest ; this explanation is supported by migration data. Late south-bound migrants tend to remain longer in the more northerly sectors of the autumn and winter range than do the early migrants. Furthermore, most of the geese banded at the Miner Sanctuary in the autumn are trapped in November and December, when the hunting season in the northern and central zones of the 1925-29 1930-34 1935-39 YEAR AFTER BANDING Fig. 77.—Comparative survival of Mississippi Valley Canada geese in four 5-year periods. Curves are based on band recoveries from geese banded at the Miner Sanctuary in the autumn (data from table 37, which include recoveries in year of banding). Curves start with an ex- pression (100 per cent) of total number of bands recovered, not total number placed on geese. Hanson & Smith: Canada Geese 925-29 n 1930-34 m 1935-39 12 1940-44 Y. 1941 -45 YEAR AFTER BANDING Fig. 78.—Comparative survival in three 5-year periods of Mississippi Valley Canada geese that were at least IJ j years old (the year after being banded). Curves I-IV are based on data from table 39, bandings at Kingsville, Ontario. Curve V is based on data from table 38, bandings at Horseshoe Lake. All curves start with an expression (100 per cent) of the total number of recovered bands that were on geese alive at the beginning of the year following banding. Mississippi flyway is at least half over. Nevertheless, these data demonstrate some- thing of the magnitude of the relative dif- ferences of survival of the various quin- quennial groupings, either graphically or expressed as survival indices. Because the Canada goose population wintering at Horseshoe Lake constitutes a somewhat different representation of the Mississippi \'alley population than do the geese banded in the autumn at the Miner Sanctuary (demonstrated by the fact that geese handed at Horseshoe Lake are shot farther north on the breeding range than are geese banded at Kingsville, Ontario, fig. 7), it is of interest to com- pare the band-recovery data from these two banding stations through the season 5-6 after banding, the last season for which data are available for both stations. When this comparison of mortality rates is made, it is desirable to omit re- coveries made during the season of band- ing, since the time of banding, the loca- tion of the banding station, and the cir- cumstances immediately following band- ing are not comparable. The recovery data from the Horseshoe Lake flock are 184 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 (li\en in table 38 and the recovery data from the Miner bandings for a comparable number of years are summarized in table 39. The survival series derived from tables 38 and 39 are presented graphically in /ig. 78. Inspection of the curves in fig. 78 reveals that the differences between curves I and IV are not so great as between comparable curves shown in fig. 77. The probable ex- planation is that all recoveries shown graphically in fig. 78 represent geese at least 11/2 years old, whereas the survival series that includes recoveries during the season of banding are in part from juvenile geese. Since the latter age class is far more vulnerable to shooting than older geese, recoveries from a banded popula- tion that includes juveniles would natural- ly reflect more sensitively the severity of hunting losses in various seasons. For this reason curve V in fig. 78, which is based on data presented in table 38, does not adequately reflect the tremendous and disproportionate kill of juveniles in the vicinity of Horseshoe Lake from 1943 through 1945. In table 40, recoveries of bandings, 1925-1939, complete through season 6-7 after banding, but omitting recoveries the 1 1925-29 n 1930-34 in 1935-39 YEAR AFTER BANDING Fig. 79.—Comparative survival (in four 5-year periods) of Canada geese that were at least VA years old (the year after being banded). Curves are based on recovery data from table 40, bandings at Kingsville, Ontario. All curves start with an expression (100 per cent) of the total number of recovered bands that were on geese alive at the beginning of the year following banding. March, 1950 Hanson & S.mith: Canad.a Geese 185 Table 40.—Recoveries in the Mississippi River valley of bands from Canada geese banded at Kingsville, Ontario, in three 5-year periods, 1925-1939. The recoveries are for the first 6 years following the year of band- ing.' 186 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 2000- 1000: 600 300H UJ > 100: < 60i ^ 30-1 m i '0: 6; 3 FROM HORSESHOE LAKE DATA •FROM KIN6SVILLE, ONTARIO, DATA T—I—1—I— r 1 2 -|—I—I—I—I—I—I— r 4 5 6 7 T—I—1—r— 9 10 -1— r 3 4 5 6 7 8 APPROXIMATE YEAR AFTER HATCHING Fig. 80.—Approximate survival curve (semilogarithmic) for the Canada goose population of the Mississippi flyway, as indicated by age ratios and censuses of geese at Horseshoe Lake, 1940- 1947, and by band recoveries from geese banded at Kingsville, Ontario, 1925-1932. Because geese banded at Kingsville were of unknown age at time of banding, the curve may be only a rough approximation of the actual survival curve. creasing age after about the fifth or sixth year of life, but the evidence is not con- clusive. The decreasing reliability of data 5 or 6 years after banding, the vary- ing take by hunters from year to year, and the fact that the data represent, in the first place, geese of unknown ages tend to obscure the actual picture. Longevity Geese as a group are noted for being long lived, particularly in captivity; Table 41.—Recoveries in the Mississippi River valley of bands from Canada geese banded at Kingsville, Ontario, in the autumn, 1925-1932. (Data from table 34.) This table diffefs from table 35 in that here the band recoveries from the year of banding are not included. 1925-1932 Bandings March, 1950 Haxson & Smith: Canada Geese 187 OS u E a o U! U na o < H a. 188 Illinois Natural History Survky Bulletin Vol. 25, Art. i Flower (1925) records that two Canada geese lived to be 29 jears of age and a third 33 years. McAfee (1924) learned of one pair of Canada geese that were mated for 42 years and another pair for over 20 years.' Wilfrid (1924) reported a gander he believed to be at least 40 years old at the time of the bird's death, and Leffingwell (1890) reported "as a matter of history" a captive bird that was killed when it was 80 years old. Doubtless tliere are other records in the literature that compare with these. Several instances of Canada geese, once used for decoys and later kept as pets, that attained ages of at least 20 years have been reported to the authors of this paper. In the wild, few Canada geese approach these ages. The greatest age attained by a wild Canada goose, to our knowledge, is at least 22 years. This goose was banded at the Miner Sanctuary in the spring of 1923 and retrapped in the spring of 1932 and again in the spring of 1944. The life span of the average wild Canada goose after banding, however, proves to be only a few years, generally less than 3, but as Austin (1942) has pointed out, "It is of little importance biologically speaking how long members of a species live providing their life span is long enough for a generation to reach and main- tain sexual maturity in order to duplicate the achievement of its predecessor." In most instances, our data are inade- quate to compute average longevities with accuracy. The complete recovery series, table 35, are of limited usefulness, since the geese involved were of unknown age when banded. These data are further complicated by the fact that the number of recoveries during the season of banding are not representative of usual first-year mortality. Average longevities calculated from data collected for the present study would be misleading. While average longevities derived from adequate data would serve ideally to compare the survival of individuals of different bandings, for the present study the survival indices shown in tables 29, 32, 40, 41, and 42 are useful and are more appropriate. From these indices and from other data, it seems obvious that few Mississippi Val- ley Canada geese live longer than 3 or 4 years after being banded. An approximation of the longevity of juvenile geese banded at Horseshoe Lake in the years of this study may be obtained through computations beginning with tiie following formula : S = fi yi 4- fi y? 4- h ya etc. N S stands for average survival after band- ing; f,, fa, etc. represent, for each age- class involved, the mortality frequency in each of successive years as computed from the survival series on page 174: 26.46, 11.53, 3.90, 3.73, 1.99 (mortalitv fre- quencies:* 73.54, 14.93, 7.63, 0.17, 1.74) ; yj, y2, etc. represent the number of years (1 through 5) following banding applica- ble to each mortality frequency ; N rep- resents the sum of the mortality fre- quencies. The mean death date of geese banded at Horseshoe Lake was about mid- way between mean banding dates. f Hence, the value calculated for S, 1.4 years, is cor- rected by subtracting 0.5 to give average survival after banding, 0.9 year. As juveniles at Horseshoe Lake were about 0.5 year old when banded, this figure is then added to 0.9 to give the average longevity, 1.4 years. Thus, it might be said that the average banded juvenile goose and presumably the aver- age juvenile in the unhanded Horseshoe Lake population in the years of this study did not live long enough to produce one brood of young. DISCUSSION It is axiomatic that the sound manage- ment of a wildlife species must in the last analysis rely on carefully gathered scientific data. Waterfowl studies usual- ly concern migratory species for which it is difficult to secure adequate data from all parts of the range. The range of most waterfowl species is immense, and seme populations shift their distribution within a flyway from year to year because of changing food, water, and weather con- ditions. The aim of most broad studies of water- fowl species probably would be to gather •Derived by subtraction: 100.00-26.46, 26.46-11.53, 11.53-3.90, etc. t Mean banding date about December 1. 1 March, 1950 Hanson & Smith : Canada Ghese 189 information that would allow manage- ment of the species concerned on a fl\ waii' basis, as recently suggested by Gabrielson (1944). Because Canada geese tend to exhibit a greater adherence to their ances- tral range than do ducks, management by riyways for this species is more suitable than it would be for most other waterfowl. In fact, the fairly restricted range of the various Canada goose populations in eastern North America, as shown earlier, suggests the need for certain management measures for indi\ idual population ranges rather than for an entire flyway. Al- though additional information concerning the Mississippi \ alle\ goose population is needed, enough is now available to per- mit this population to be managed pri- marily as indi\idual population units. Status In 1946, 14 states of the Mississippi flnvay (Michigan, AVisconsin, Minnesota, Ohio, Indiana, Illinois, Iowa, Kentucky, Tennessee, Missouri, Mississippi, Ala- bama, Arkansas, and Louisiana) were closed to the hunting of Canada geese. The closed season of an entire flyway was the first of its kind in the history of this species of waterfowl. The only similar actions ever taken were those closing the shooting seasons on snow geese and brant in the Atlantic Coast states. Snow-goose hunting has been prohibited there since 1931, and brant hunting for more than half of the years since 1933. The closed season on Canada geese in the Mississippi flyway in 1946 was belie\ed necessary for a number of reasons: an alarming decrease in the number of these geese in the Mississippi flyway from 1940 to 1945, as indicated by January inven- tory data; markedly increased kills be- ginning in 1939, particularly in the region of Horseshoe Lake ; a disproportionate kill of juvenile birds and an apparent de- creased productivity in 1945, as indicated by research at Horseshoe Lake. The peak number of geese at Horseshoe Lake dropped from about 50,000 in 1943- 44 to 26,000 in 1945-46. That this de- crease represented a real decrease in the flyway population and was not due to by- passing of the area by flocks is shown not only by flyway censuses but b\- band-re- covery records. These records indicate that since 1932 many of the geese that formerly used the Mississippi River from Cairo, Illinois, to Baton Rouge, Louisiana, have concentrated in a much smaller area centering on Horseshoe Lake, probably because of the refuge there and the large amount of grain available to the geese. Known bags and careful estimates of kills indicate that, in the years just pre- vious to 1946, an average of about 20 per cent of the Canada goose population w intering at Horseshoe Lake was bagged annually, and that the total annual kill in the area averaged about 27 per cent of the population. In view of the fairl\ low productivity of the Canada goose, it is obvious that a reasonable kill in this area was greatly exceeded. Population declines at Horseshoe Lake and in the Mississippi flyway as a whole showed that flock mortality from all causes combined had been excessive, and, as hunting losses are one type of mortality that can be con- trolled, it was evident that closing the en- tire flyway to shooting was the most effective management measure that could have been employed. Evidence of increased shooting pressure on Canada geese in years just previous to 1946 is illustrated by the survival curves, fig. 77, representing data computed from band recoveries from geese banded at the Jack Miner Bird Sanctuary, Kingsville, Ontario. These data show that the an- nual survival rate for that portion of the population migrating through the Kings- ville area was lower in the 5 years be- ginning in the fall of 1940 than in any comparable period in the previous 15 years, fig. 77. Chiefly responsible for this lower survival rate were the heavy kills made at Horseshoe Lake ; band recoveries show that the survival rate of the Horseshoe Lake flock was well below the average for the entire Mississippi Vallev' population. In fact, the survival series for the Horse- shoe Lake flock was lower during the period 194(J-1945 than it was in the entire Mississippi Valley population in the years in which baiting and the use of live decoys were permitted, tables 10, 37, and 38. Moffitt (1935) was concerned over the future of a flock nesting in California when he realized an 11.5 per cent first- season recovery rate from his bandings. Unpublished studies by Cecil S. Williams 190 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 of the United States Fish and Wildlife Service at the Bear River marshes, Utah, indicate that the Great Basin population he dealt with could show a first-year band- recovery rate of 16 per cent and a total band-recovery rate of 25 per cent and still increase. Total recoveries from Horseshoe Lake bandings were at only about half the rate of total recoveries re- ported for the Great Basin, but other data indicated a heavy kill rate and a decline in the Horseshoe Lake population in the years just previous to 1946. While Wil- liams' data establish the fact that the Canada goose could withstand heavy shooting losses in the Great Basin, con- ditions vary too widely in the various fly- ways to predict on the basis of data from one area (Utah) what the conditions are in another (Horseshoe Lake). Interpolating from fig. 74, curve I, which is based on a survival series ob- tained for the Horseshoe Lake flock, it appears that only about 16 per cent of the juveniles reaching Horseshoe Lake during the period of field work for this study lived long enough to see a brood of their young on the wing. When a major portion of the annual kill of a Canada goose flock is at the ex- pense of one age group, data on the total number of birds bagged do not reveal the true impact of the kill upon the total population. At Horseshoe Lake the juve- niles made up the major part of the kill in the period covered by this study, tables 43 and 44. In the autumn of 1943, the juveniles made up 56 per cent of the population, while 91 per cent of the hunt- er's bag consisted of juveniles. In that year, 37 per cent of the juvenile popula- tion at Horseshoe Lake was bagged. The following year, the 1943 generation (then yearlings) comprised only about 29 per cent of the total adult birds. The effect of this differential kill is also shown by trap-age ratios of banded survivors in later years. In table 43, returns for the years 1943-1947 of geese banded during the autumn and winter season previous to Table 43.—Juvenile-adult ratios of Canada geese at Horseshoe Lake, 1942-43 through 1946-47, xMarch, 1950 Hanson & S.Mr]'>i: Canada Geese 191 each of these years are given. In the autumn of 1942, when only 2.59 juveniles were shot for every adult, the survival rate of juveniles was evidently favorable to this age class as 10.82 per cent of the total banded juveniles returned to the traps in the following year as compared with 7.92 per cent of the adults, or a ratio of 1.37 juveniles to 1.0 adult. In contrast to this survi\al picture is the highly diiierential kill that occurred in 1943 when the ratio of juveniles to adults shot at the hunting clubs surround- ing the refuge was 10.53 to 1.0, table 43. The next year the return to the traps was only 0.56 juvenile (then yearling) to 1 adult. Despite the fact that juveniles bore the brunt of the kill in 1944, table 44, the net loss to the juvenile segment of the population was somewhat less, with the result that the ratio of juvenile ( then yearling) to adult returns in the traps a vear later, in 1945-46, was 1.73 to l.d, table 43. A relativel.\' higli kill of juveniles coupled with a .\ear in which productivity is low is almost certain to place a goose population in a hazardous position. Band- ing at Horseshoe Lake indicated a de- crease in productivity in 1945 from the productivity in 1944, table 43. The rel- atively small number of young produced in 1945 may have been related in part to the cold weather in the spring of that year ; the productivity of mallards also was greatly reduced in that year. A depres- sive effect on the intensity of mating or on nesting success in many species of birds has been attributed to late and cold springs. The following species said to be affected thus might be cited : Canada goose (Johnson 1947) ; arctic tern (Lack 1933); eiders and loons (Bird & Bird 1940); moor hen (Huxley 1932); and house wren (Kendeigh 1942). However, it is conceivable that part of the decrease in productivity in 1945 may have been apparent rather than real. Un- doubtedly juveniles contribute a larger proportion of the kill during migration than do the adults, but the extent to which shooting north of the Horseshoe Lake Game Refuge is selective of juveniles is not known. Because the kill between the refuge and the Canadian border in 1945 was much larger than usual, the age ratios in the southward-bound flocks may have been significant!) altered by shooting in that particular year. The subject of cycles in waterfowl is still largely an unexplored field. It does not appear to be known generally that, in the arctic, geese, ducks, and loons nuu be subject to nonbreeding years (Man- niche 1910, Bertram, Lack, c^ Roberts U)34, Bird S: Bird 1940). Keith (1Q37| writes, ". . . . 1936 was a 'non-breeding year' [in Northeast Land, Spitzbergen Archipelago] when large numbers of Ducks and Geese failed to nest; and in other parts of the Arctic it had always before been found that the Divers [loons] were also affected by these years and that of them too only a small proportion were breeding." As nonbreeding of v\aterfowl has been reported only from high arctic areas, it is debatable whether the Canada goose populations dealt with here are similarly affected. At present, low productivity in blue geese and snow geese appears to be con- fined to summers in which inclement weather directly affects the success of nesting (Soper 1930). In the opinion of Berry (1939), "climate is of the utmost importance in limiting the survival rate of goslings on the northern breeding grounds." A year of low productivity in Canada geese should be of particular concern to the administrators who seek to influence the kill by hunting regulations, for the reason that the young birds bear a double responsibility. Being more vulnerable to shooting than the adults, they must con- tribute a disproportionate share of the kill, and, secondly, they must survive in sufficient numbers to help reproduce an equivalent of the annual loss in the breed- ing population. Even in a year wlien the pioduction of young was not signilicantl\ low, 1943, shooting losses in the Horse- shoe Lake area were so severe and so greatly at the expense of the juveniles that only a small proportion of this generation survived to reach the minimum breeding age of 2 years. Management What can be done to insure the future of the Mississippi Valley geese? Lentil recent years, two prime measures for con- 192 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 serving waterfowl, hunting regulations and refuges, have been fairly successfully used in the management of this group. As applied to the population of geese dealt with in this report, it is apparent that these measures were not very effective in the period of field work. In Canada.—Several factors minimize the need for any immediate change in measures relating to the Mississippi \^alley population while in Canada. The relatively inaccessible nature of the Cana- dian breeding grounds insures adequate protection for the flock during the actual breeding season. The kill in Canada is not excessive, and a reduction of the early spring kill on the breeding grounds would be difficult be- cause much of this kill is virtually neces- sary for the survival of native Indians. Furthermore, our kill and population data indicate that the annual rate of kill (the percentage of birds taken from the re- turning population in the spring) by the Indians is relatively constant. In general, only when there is an actual increase in the Canada goose population does a signif- icant increase in the number of these geese bagged by Indians occur. This relatively constant relationship is evidence that the goose kill by natives cannot be considered the direct cause of any considerable popula- tion decrease that might be reported in the United States from any of the annual January inventories. In the United States.—In 1944 and 1945, when season bag limits were im- posed for Alexander County, Illinois, table 10, it was a relatively easy matter to limit the kill of geese in the Horseshoe Lake area to approximately the predeter- mined figures. The facility with which the day-to-day kill can be tallied is perhaps the outstanding advantage of en- couraging a portion of the flock to utilize the refuge there. The season bag limit in the above instances was determined by the trend of the population in prior years, but, to be fully effective, management should anticipate future trends based upon the current composition of the population. With the data at hand on the Canada Table 45.—Calculated losses and reproductive gains for the Horseshoe Lake Canada goose flock between the autumn of 1944, and the autumn of 1945.' Classification March, 1950 Hanson & Smith: Canada Geese IW goose in the Mississippi River valley, it is possible to arrive at a practical estimate of the maximum kill that can be tolerated. A method b\ which management of the Mississippi \'alley population might pro- ceed is best illustrated by a concrete example, table 45. Similar calculations based on sex and age ratios from trapping, and made b>' the authors in the spring of 1945, forecast a decreased population for the autumn of 1945. Censuses during the autumn and the inventory of Januarj', 1946, proved the accuracy of this pre- diction. Since the autumn flight in any year depends to a large extent on the produc- tion of young in the spring of that \ear, it is necessary to know the approximate number of breeding females and to have some measure of the nesting success on the breeding grounds to predict the autumn flock population with reasonable accuracy. In\entory on the breeding grounds would be difficult because of the nature of the terrain, but the use of planes would aid tremendously in such work. For the present, and until more data are available, the average produc- tivity of the population might be calcu- lated on the basis of three young (brought to flying stage) per adult female. If the flock population has been fairlj' stable for several years or is on the in- crease, a bag of 10 per cent of the number wintering in the Horseshoe Lake area might prove to be within the limits of what the flock could stand without decreasing in size. Even this kill might be too high if kills north of Horseshoe Lake were un- usually large in a gi\en autumn, if nesting success was low the previous spring, or if sex and age ratios were seriously un- balanced. When the population is \ery low, the kill of a single bird constitutes overshooting. A reduction in the crippling loss would allow the season bag limit in the Horse- shoe Lake area to be increased. The num- ber of geese crippled and lost to hunters each year in the area is needlessly high. An estimate of cripples not retrieved and soon dying is placed at 30 per cent of the number of geese bagged. Certain ad- ministrative measures can be taken to re- duce the per cent of cripples not retrieved. For instance, adequate spacing of pits to reduce competition among hunters would materially aid in reducing crippling losses. But a large share of the responsibility will rest with the hunter himself, who must restrain the natural desire to "give a high one a ride." Some hunters hope to bag geese with greater ease by using magnum shotguns. However, it is open to debate wiiether more geese are bagged than crippled by such guns because of the out- of-range shooting their possession en- courages. At least in one instance a 10- gauge magnum shotgun is known to have failed to li\e up to its owner's expectation ; a tally of empty casings from this shot- gun in one pit, presumably fired to bag the limit of two geese, was 22, as against the average of 9 cartridge casings per hunter for all pits inspected. It is clear from tables 15 and 10, show- ing kill and hunting regulations in the Horseshoe Lake area, that hunting restric- tions were not always successful in re- ducing the kill to the desired extent, but. if various measures instituted to lower the annual kill had not been taken, it is probable that a large proportion of the Canada geese using the Horseshoe Lake area would have been shot by the end of 1945. Under normal conditions, the duration of the hunting season can be expected to show a fairly direct relationship to the kill, but, when the natural wariness of the geese has been reduced, as at Horseshoe Lake, the length of the hunting season may show no correlation with the kill, figs. 52 and 53. Pirnie (1939) has emphasized that "Changing habits of these birds [Canada geese] may create new hazards for them and reipu're even more stringent regula- tions." The behavior of the Horseshoe Lake flock in recent years and its relation to shooting has already been discussed, but it should again be emphasized that restric- tions alone cannot be expected to safe- guard it. Refuges form an important part of our system for the preservation of waterfowl. Whether or not any individual refuge proves of value will depend to a certain extent upon its management and also upon its size. Leopold (1931) stated the chief problem in regard to the Horseshoe Lake Game Refuge soon after this refuge was 194 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 created. "The question of whether public refuges should be surrounded by public shooting grounds is frequently debated. Horseshoe Lake in Alexander County, Illinois, is a good place to study the ques- tion." Twelve years after this statement was published the answer was forcibly given by Gabrielson (1943). "Because of its [Horseshoe Lake Game Refuge] attrac- tiveness to Canada geese, small size, lack of food, and peculiar relation to surround- ing lands, it has become a slaughter pen rather than a refuge." The breakdown in wariness that oc- curred was perhaps more serious to the future of the Horseshoe Lake flock than the reduction in its size. The steps be- lieved necessary to re-establish wildness in the flock were as follows : ( 1 ) Establish refuge areas on the nearby islands and bars of the Mississippi River or on lands adjacent to the river; (2) disperse the geese from Horseshoe Lake to these bars and islands ; that is, drive them back to their original habitat; (3) insofar as possible, reduce contact between human beings (both the public and refuge per- sonnel) and the geese. In the past years in which the geese used both the river bars and the refuge, they retained their natural wildness; coin- cident with their almost complete depend- ence on the refuge for food and grit, they lost much of their wildness. The river refuge might act as a final sanctuary for the flock should it be disturbed for any reason at Horseshoe Lake, and ideally it should contain the bulk of the flock at most times. Canada geese will feed by moonlight, at daybreak, or at dusk, if they are disturlsed while feeding during the day. This fact may offer a partial solution to the Horse- shoe Lake problem. If the geese were permitted to feed at the Horseshoe Lake Game Refuge only during the hours of dawn and dusk, the re-establishment of wildness might occur and with it a reduc- tion in the rate of kill. We have a prece- dent for such a course of action in the operation of the Miner Sanctuary, where the geese feed only in the early morning hours and at dusk, spending the remainder of their time roosting out on Lake Erie. State regulations just previous to 1946 prohibited the placing of shooting pits within 75 to 150 yards of the boundary of the Horseshoe Lake Game Refuge. This buffer zone, which was intended to allow the geese to attain safe heights before reaching the shooting pits and blinds was unquestionably insufficient, since many of the geese leaving the refuge encountered shot pellets 75 yards away from the first line of pits. Although the Miner Sanc- tuary consists of only 400 acres and sup- ports an even greater density of geese than is ever experienced at Horseshoe Lake, excessive kills have not occurred near this Canadian refuge in late years. Responsi- ble in part for the small kills reported in the vicinity of the Miner Sanctuary is a buffer zone that surrounds the ponds and feeding grounds for a distance of a mile. When geese leave the refuge, they have sufficient space in which to gain altitude before passing over the shooting grounds. The present food resources of the Horseshoe Lake Game Refuge are insuffi- cient to winter more than 20,000 geese, and probably only 15,000 can be accom- modated to best advantage. When the corn crop and wheat brovvse on the re- fuge are exhausted, and sometimes before this occurs, the flock feeds on unharvested and waste grain and on the green plants of winter wheat in fields of the surround- ing countryside—occasionally at a con- siderable loss to farmers who do not rent their fields to hunters. Unless the flock is broken up and scattered to other areas in the flyway, the local food conditions must be improved, either through the ac- quisition of more land or by an artificial feeding program. The artificial program is wholly undesirable unless it is carried out on an isolated tract of land. On the other hand, the development of a river refuge would certainly increase the flock's usage of natural foods—the grasses, sedges, and switch willows on which the geese formerly fed. The present size of the Horseshoe Lake Game Refuge is woefully inadequate for the geese using the area, as experiences there and elsewhere have demonstrated. A program involving purchase of additional lands has been planned by the State De- partment of Conservation for several years, but has been blocked by the in- flated prices of lands in the area—inflated March, 1950 Hanson & Smith : Canada Geese 195 prices resulting in part from the commer- cialization of goose hunting. Census data showed that, between 1942 and 1945, the Canada goose in the Mis- sissippi River valley suffered a marked decline in population. Kill records showed an increase in the annual bag beginning in 1939, and banding data revealed a con- current decrease in goose survival for the same period. The conclusion must be reached that the Mississippi Valley Canada goose population was shot too heavily in that period and that stringent protection was necessary to insure perpet- uation of this population. PRESENT SITUATION The time lapse between completion of the field work reported here and publica- tion of this article has been sufficient to permit an evaluation of some of the meas- ures recently taken to assure the future of the Canada goose population of the Mississippi River valley. The decision to close the valley to Canada goose hunting in 1946 was based partly on evidence gained from banding that the geese winter- ing at the Horseshoe Lake Game Refuge were suffering unprecedented losses from hunting and were being killed at a rate far greater than the flock could stand and still maintain its numbers. In addition was the evidence from annual inventories that the flyway population was at an alarmingly low level. In 1947, the shooting of Canada geese was again permitted in the Mississippi River valley, but on a restricted scale. The season opened on November 4 and closed on December 3. The bag limit was reduced to one bird per day and the possession limit was also one bird. To in- sure against a return of heavy kills in the Horseshoe Lake region, an area in the region totaling approximately 15,000 acres was declared closed by proclamation of the President of the United States with the joint support of the Governor of Illinois. By this action, a buffer area, roughly 2 miles in depth, was created around the Horseshoe Lake Game Ref- uge. In 1948, the hunting season opened on October 29 and closed November 27. During this 30-day season, hunters were permitted to bag two Canada geese per da.\- and were allowed a possession limit of two birds. The response by the geese to greater protection has been most heartening, their comeback demonstrating both that the kill by hunters in the United States was a major suppressive factor on the popula- tion, and that this population, given op- portunity, possesses strong recuperative powers. ^Vith a capital investment of 49,000* birds in the I\Iississippi flyway in the winter of 1945-46, interest in the form of 1946 reproduction was reinvested as capital gain by virtue of the closed season. Inventory in January, 1947, re- vealed a capital gain of approximately 25 per cent, table 7. This reco\ery by an almost bankrupt population so encouraged the committee on regulations that a di\idend, in the form of an open season, was declared permissible for the autumn of 1947 and again for the autumn of 1948. The dividend in the Horseshoe Lake area in 1947 was 1,644 geese bagged by hunt- ers; in 1948 it was 2,587 geese bagged by hunters. In addition to this number, other geese, estimated at 2 000, were shot illegally within the buffer area closed to hunting outside the refuge. We do not have the data at hand to show what the profits were to hunters in other states in the flyway, but that the goose business could afford the dividends is shown by the recent summar\' of capital stock given in table 46. The recovery made by the Mississippi \'alley population has not gone unnoticed by the Indians who trap and hunt on the breeding grounds before the actual com- mencement of nesting. In August, 1949, the senior author learned at Fort Albany that the Indians there had observed more geese in the spring of 1949 than at any other time in recent years. Similarly, questionnaire answers received from Ray- mond M. Alaine of AVeenusk, September 21, 1949, stated that the Indians at that post had not seen as many geese in any other years of the last 10 as thev did in the fall of 1948. Future management of the Horseshoe Lake flock bv the United States Fish and ' This figure indudes geese from western _ Louisiana, birds that possibly belong to the Eastern Prairie popula- tion and tiiat should not be included in the Mississippi Valley population. Hence, it exaggerates the size of the Mississippi Valley population for 1945-46. 195 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Table 46.—Population of Canada geese in the Mississippi River valley in 1947-48 and 1948-49, from January inventory, except as noted. Season State or Other Area Michigan. . , Wisconsin Indiana Illinois (total) Mason County Horseshoe Lake Craborchard Lake, William- son County Lyerly Lake, Union County Miscellaneous areas Kentucky Mississippi Tennessee -Arkansas Missouri Louisiana Total 30,000 12,450 1,004 7,200 7,250 9,450 12,000 5,000* 10,000 159,523 * Most of these geese were a part of tfie Rock wintering in the Horseshoe Lake area. Wildlife Service and the Illinois Depart- ment of Conservation envisions the break- ing up of this concentration and dividing it among four other refuge areas: Crab- orchard National Wildlife Refuge, Wil- liamson Count}', Illinois; Lyerly Lake State Refuge and Public Shooting Grounds, Union County, Illinois ; the Mingo National Wildlife Refuge, Mis- souri ; and the Kentucky Woodlands National Wildlife Refuge bordering the Tennessee River south of Paducah, Ken- tucky. To implement the dispersal of geese from Horseshoe Lake, planes, guns, bombs and various other pyrotechnic de- vices were used to frighten the geese in 1947, 1948, and 1949. That this dis- persal program is meeting with success is evident from the data presented in table 46. Provided with these other areas, an ample food supply, and adequate legal pro- tection, the Canada goose population in the Mississippi valley faces a future that seems assured for some years to come. SUMMARY \. The Horseshoe Lake Wildlife Ref- uge, located at the southern tip of Illinois near Cairo and created in 1927 by the Illinois Department of Conservation, totals approximately 3,700 acres. 2. Soon after the refuge was formed, increasing numbers of Canada geese, de- coyed from their traditional wintering grounds along the Mississippi River by the food and protection offered, began to use this refuge. In most recent winters the refuge has attracted about 50 per cent of the Mississippi Valley Canada goose population. With the increase in the size of the flock at the refuge, there was a loss of wariness on the part of the geese, accompanied by a tremendous increase in the annual kill. 3. In the eastern half of the United States there are two subspecies of Canada geese. The easternmost race, Branta canadensis canadensis, comprises the geese of the North Atlantic population. The other race, Branta canadensis interior, which breeds principally west, south, and east of James and Hudson bays, is com- posed of four subgroups, each of which constitutes a separate flyway population. The four subgroups are as follows : the South Atlantic, the Southeast, the Mis- sissippi Valley, and the Eastern Prairie. 4. The main breeding range of the Mississippi Valley geese is believed to lie within the western limits of the Paleozoic Basin west of James Bay and south of Hudson Bay. The majority of the nest- ing geese of this population are found in relatively restricted areas of the vast, low- lying, muskeg-covered plain of the region. 5. Aerial observations revealed that the type of muskeg attracting the greatest numbers of geese is one that is studded with potholes of a few acres to about 30 acres in size, so closely grouped that often only a narrow strip of land or floating vegetation separates one from another. 6. Most nesting pairs of Mississippi Valley geese are concentrated in produc- tion centers, but, as most of these produc- tion centers are of considerable size, scattered nesting, with one or two pairs to a small lake, seems to be the rule west of James Bay and south of Hudson Bay. 7. Before the southward migration of Mississippi Valley geese begins, about August 15, some family groups and small flocks begin a series of local flights, the termini of which are favored feeding grounds along the west coast of James Bay and the south coast of Hudson Bay, the tundra of Cape Henrietta Maria and the coastal marsh of Akimiski Island. The March, 1950 Hansox & Smith : Canada Geese 197 tundra of Cape Henrietta Maria is favored because of the quantity of berries usually available there. 8. At least half of the Alississippi Valley geese do not fly to the coastal areas before migrating, but lea\e directly from their muskeg breeding grounds and strike south on a broad front. These are be- lieved to be the geese that cross the Canadian border into eastern Minnesota and the upper peninsula of Michigan. 9. While probably at least a few Canada geese in migration pass over most areas of the Mississippi flyway each year, band recoveries and observations indicate that the following routes are most fre- quently used : from the Miner Sanctuary to Horseshoe Lake via Lake St. Mary, the Wabash and Ohio rivers; from Saginaw Bay southwest across the lower peninsula of Michigan to the W. K. Kellogg Bird Sanctuary area and the lower Kalamazoo River; southward along both shores of Lake Michigan. Migration through Wis- consin is principally in the eastern half of the state. The west shore of Lake Michi- gan is followed by appreciable numbers of geese. Two other routes appear to be favored : ( 1 ) the valley of the Wisconsin River; (2) Green Bay south to Lake VV'innebago, the flight probably splitting south of Lake AVinnebago, one section going to the Lake Geneva area, the other following the Rock River valley. Migra- tion through Illinois appears to take place on a fairly broad front although the Illinois River valley is particularly favored. 10. Band-recovery data indicate that turnover in the population wintering at Horseshoe Lake is negligible. Geese that are decoyed into this refuge usually re- main there for the rest of the season. 11. A portion of the Mississippi Valley geese migrating through the Kingsville, Ontario, region do not visit the Horse- shoe Lake Refuge but by-pass it to the east, probably via the Tennessee River, and winter on the lower Mississippi. 12. The northward migration in spring is more nearly on a directly north and south axis than routes taken in the autumn. The flights of Mississippi V^alley geese that stop at the Miner Sanc- tuary in the autumn do not reappear there in the spring in appreciable num- bers; presumably they return to the breeding grounds by a more westerly route. The spring flights through the Kingsville region are comprised chiefly of South Atlantic geese. 13. Autumn migration of Mississippi Valley geese occurs over a 3-month period ; the last geese to reach Horseshoe Lake in appreciable numbers arrive in early December. Much of the late flight rep- resents the exodus of geese from the Miner Sanctuary when feeding there is curtailed. 14. The southward movement of the Canada geese from the breeding grounds may be compared with a segment of the concentric waves produced by an object striking the surface of a body of water. Geese that leave the breeding grounds earliest are believed to winter in the most southerly areas of the flyway. Those that leave the breeding grounds last are be- lieved to winter in the most northerly areas of the wintering grounds. 15. In spring, the first flocks generally arrive on the breeding grounds between April 15 and 25, 2 to 3 weeks before the breakup of the major rivers. 16. Winter concentrations of Canada geese occur in the region of Kingsville, Ontario, westward to southern Wisconsin, and south to the Gulf Coast. 17. Although the Canada goose is widely reputed to be an extremely wary and difiicult species to hunt, the behavior of this species at Horseshoe Lake in re- cent years has contradicted this reputa- tion. Believed responsible for the high vulnerability of Canada geese to shooting in the vicinity of this refuge are the psychologically pacifying effect of large numbers of geese at rest on a relatively small area ; the frequent sight of man in a benign role; and the decreased mobility of the flock when food is abundant on the refuge, as well as on adjacent hunting areas. 18. Goose hunting in Illinois, once a sport carried out in widely scattered areas of the state, is now confined largely to the Illinois River valley and the Horseshoe Lake region. 19. In the period 1944 through 1947, the kill of Canada geese of the Mis- sissippi flyway by Canadian Indians is computed to have been between 4,000 and 5,500 or from about 8 to 10 per cent of the number of birds that attained the breeding grounds in the spring. Approxi- 198 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 mately 25 per cent of the total number of Mississippi Valley Canada geese bagged in recent years have been taken by Indians. 20. The waterfowl kill made by the Indians of the James Bay region is some- times vital to actual survival of the In- dians. Blue geese and snow geese greatly outrank the Canada goose in importance during the fall hunt along the coastal marshes; in spring, when the Indians are trapping inland along the rivers and creeks, the principal kill of Canada geese occurs, while relatively few blue geese and snow geese are shot at this time. 21. The restocking of beaver in some areas of the Canadian goose breeding range is beginning to relieve some of the hunting pressure on Canada geese. 22. The kill in the Horseshoe Lake area first began greatly to exceed what the flock could stand in 1939 when a kill of 17,300 geese was made. The average number of geese bagged in the Horseshoe Lake area in the autumns of 1939 through 1945 was about 9,800. In the autumns of 1943, 1944, and 1945 the bag amounted to 23, 19, and 18 per cent, respectively, of the number of geese that arrived at the refuge in those years. 23. The annual bag of geese in Illinois in areas other than Horseshoe Lake averaged approximately 1,100 birds in the period covered by this report. 24. Next to Illinois, Michigan made the largest kills of Mississippi Valley geese, 1938-1944; the annual bag was probably between 1,000 and 3,000 birds. 25. Bag inspections at hunting clubs near Horseshoe Lake showed that juvenile geese made up a high percentage of the total kill, 1940-1945. In 1943, juveniles were about eight times as vulnerable to hunting as adults. 26. Crippling losses among geese at Horseshoe Lake in recent j'ears are esti- mated to have been equivalent to about 30 per cent of the annual bag. 27. Causes of death among Canada geese at Horseshoe Lake include lead poisoning (from ingestion of lead pellets), bound crop (perhaps a result of lead poisoning), tracheitis, and aspergillosis. 28. Sex ratios obtained from trapping geese at Horseshoe Lake show that there were slight, but statistically significant, larger numbers of males than of females in the juvenile and adult age classes, 1940-1946. Bag-inspection figures showed no significant preponderance of either sex in either age class, 1940-1945. 29. Nesting success of geese is not ap- preciably affected by the Canadian Indians, since the bulk of the kill is made in early spring before geese have begun to nest. Foxes may have a slight effect on nesting success when their other prey species, which appear to be cj'clic, are low in numbers. 30. In 7 years of trapping and bag inspection at Horseshoe Lake, the age ratios obtained varied from 57 to 204 juveniles per 100 older geese. In 1944- 45, trapping indicated that 55 per cent of the population consisted of juveniles. Trapping in the following year indicated that the proportion of juveniles had dropped to 36 per cent. 31. Average flock size, computed from frequency counts of flocks of nine or fewer geese on the wintering grounds, may provide a quick means of appraising breeding success of geese in the previous spring. 32. Low survival of Canada geese banded at the Jack Miner Bird Sanc- tuary, 1940-1944, is believed to have been brought about chiefly by the tremendous increase in the kills made in the region of Horseshoe Lake. 33. Mortality data calculated from trap- ping and band-recovery figures show that the Horseshoe Lake flock had a lower sur- vival rate during the period of this study than did comparable banding classes from the Miner Sanctuary. 34. Mortality indices, the average of mortality rates for three years after band- ing, provide a possible basis for com- paring mortality between different popu- lations and banding classes of geese. 35. Survival data for the Horseshoe Lake flock, 1941-1945, indicate that the average juvenile did not live long enough to produce a brood of young. 36. In 1946, no open hunting season on Canada geese was permitted in the Mississippi River valley. In 1947, shoot- ing on a restricted scale was permitted. 37. Increased protection of the Mis- sissippi Valley Canada geese plus certain other management practices resulted in an appreciable gain in the population by the fall and winter of 1948-49. APPENDIX A THE SOUTHEAST POPULATION ONE of the important findings from our study of the Jack Miner banding data, as they rehite to the Horseshoe Lake prob- lems, is the existence of a distinct and here- tofore unrecognized group of Canada geese that winter in the inland areas of Virginia, North Carolina, South Carolina, Georgia, and Alabama and on the Gulf Coast of Florida. Because management of the Mis- sissippi V^alley goose population should be guided to some extent by a knowledge of neighboring goose populations, it seems desirable to include in this paper a brief summary of the breeding and wintering ranges, as well as the migration paths, of the Canada geese of the newly defined group, to which we have given the name Southeast population. Breeding Range To date there have been no recoveries of Horseshoe Lake goose bands in the Moose River district of James Bay or at the extreme south end of this bay, while fair numbers of bands have been recovered in that region from geese banded at the Jack Miner Bird Sanctuary near Kingsville, Ontario. Large numbers of Miner bands from the autumn flight have been recovered in the inland portions of the southeastern states. It appears from band recoveries that the Southeast geese nest from the coun- try drained by the Moose River, south and east to the Nottaway or Rupert river country. In an area north of the Moose River, the breeding grounds of these geese merge with those of the Mississippi Valley population; east of the Nottaway River, or Rupert River, they merge with the nesting grounds of the South Atlantic geese, most of which migrate through the Kingsville area only in spring. Census data on the flyway of the South- east population are meager. Because the scattered flocks were not recognized as components of this distinct population, their significance was lost in the usual method of lumping census figures by states. Popula- tion figures presented below are from three sources: letters to Jack Miner from local sportsmen or officials; personal conversation with W. P. Baldwin, Jr., United States Fish and Wildlife Service biologist, stationed at Port Wentworth, Georgia ; and records in the files of the Division of Refuges, United States Fish and Wildlife Service. Following is a summary of the wintering grounds of the Southeast population, as in- dicated by band recoveries and other data. Migration Routes In the autumn migration, the range of the Southeast population overlaps that of the Mississippi Valley population between James Bay and the Miner Sanctuary. At the latter point, however, band recoveries indi- cate that the birds of the Southeast popula- tion split off from the Mississippi Valley population and fan out south and southeast over a number of courses. The paucity of recoveries between the Miner Sanctuary and the eastern and southern slopes of the Appalachian Mountains suggests that most of the geese of the Southeast population make few stops en route to their wintering quarters. The routes taken by these geese on their northward migration are probably mainly to the west of their autumn migration paths, as band recoveries show that comparatively few of the birds retrace their autumn flight through the Kingsville, Ontario, region. Winter Concentrations The wintering grounds of the geese of the Southeast population lie mainly in the Piedmont region east and south of the Appalachian Mountains, and in some parts of the coastal plain. The wintering range can be better understood if the distribution of the recoveries from the southeastern states in figs. 12-21 is compared with the physiographic features of these states shown in fig. 81. Band recoveries show that geese resort to nearly every river of appreciable size that dissects the Piedmont and the coastal plain, but that the numerous reser- voirs are particularly favored. The coastal [ 199 200 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 / { March, 1950 Hanson & Smith: Canada Geese 201 along the Great Pee Dee River and at Ansonville has recently totaled about 3,000 (personal communication, March, 1949). South Carolina.—Pickens (1928) re- ported the Canada goose to be a common winter resident in upper South Carolina, a statement that is amply supported by the Miner returns of autunui-banded geese. According to Ernest F. Holland, manager of the Carolina Sandhills National Wildlife Refuge (letter to the Jack Miner Bird Sanctuary, December 18, 1946), about 5.000 Canada geese used this refuge and the adjacent Great Pee Dee River basin in the late autumn of 1945. An additional 2,500 were reported using the private waterfowl refuge of Lockhart Gaddy, lo- cated near Ansonville, North Carolina. Judged from band recoveries, Lake Murray, an impoundment of the Congaree River, is probably one of the more important bodies of water for Canada geese in South Caro- lina. Wateree Pond, a much smaller res- ervoir on the Wateree River, appears to be second in importance. Other rivers used by geese are the Broad, the Saluda (Lake Greenwood), and the Savannah (from Anderson to Aiken counties). In several years prior to the winter of 1948—1-9, about 2(X) Canada geese wintered in the vicinity of McBee in Chesterfield County and about 250 on Lake Murray (W. P. Baldwin, Jr., personal communi- cation, March, 1949). Another 300 fre- quented the section of the Savannah River bordering McCormick County. The Santee Cooper Reservoir area harbored about 250 Canada geese, the Cape Romain sector of the Atlantic Coast about 500, and Winyah Bay a small but unknown number. The Winyah Bay flock may be only a segment of the Cape Romain flock that segregates out from time to time. Data on populations at the Santee Cooper and Cape Romain National Wildlife refuges in other recent years are given in table 47. Alabama.—^According to Howell (1924), Canada geese in Alabama are "probably most abundant on the Tennessee River in the vicinity of Muscle Shoals." They are "numerous every winter in the vicinity of Montgoinery. On the coast they apparently are not common, though found occasional- ly." Since 1942, 300 to 900 Canada geese have been reported wintering in the vicinity of the Wheeler Reservoir, according to data in the files of the LTnited States Fish and Wildlife Service. Sixty per cent of the Miner-banded geese reported killed in Alabama were shot in Tallapoosa, Coosa, and Elmore counties. Over half of the recoveries from these three counties are from the vicinity of Mar- tin Lake, an impoundment of the Tallapoosa River; the remaining returns from these counties are from areas adjoining the Coosa Table 47.—Numbers of Canada geese wintering at three national wildlife refuges in the Southeast flyway, 1934-1945. 202 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 River impoundments (Lay, Mitchell, and Jordan lakes). The flock wintering on Martin Lake numbered about 400 in the winter of 1939-40 (letter to Jack Miner from C. Robinson of Alexander City, Ala- bama). Georgia.—Band recoveries from Georgia are spotty, suggesting that no great con- centrations of geese occur anywhere in the state, possibly in part because of the com- parative lack of large reservoirs or natural lakes. The Savannah River from Hart County to Richmond County appears to be a favored wintering area; the Ocmulgee (Lloyd Shoals Reservoir between Jasper and Butts counties), the Oconee (Washing- ton and Laurens counties), and the Flint River (Pike, Upson, Taylor, and Craw- ford counties) are other sectors used by Canada geese. Many of the recoveries from Georgia, however, may be from migrating geese rather than from wintering flocks. As a number of the recoveries are from areas of the state directly north of the St. Marks National Wildlife Refuge on the Florida Gulf Coast, it seems likely that birds en route to St. Marks contribute appreciably to the kills made in Georgia. In 1941, 150 to 200 Canada geese were reported using Lake Harding, an impound- ment created by Bartletts Ferry dam on the Chattahoochee River near West Point (letter to Jack Miner from William B. Fuller, West Point, Georgia, January 10, 1941). Florida.—The St. Marks National Wild- life Refuge, consisting of 54,681 acres, is believed to contain the greatest single con- centration of geese in the Southeast popu- lation. Although between 11,000 and 15,000 geese have wintered at this refuge since 1941, table 46, there have been singularly few band recoveries from Miner-banded geese in the surrounding country. This fact suggests that either the bulk of these geese by-pass the Miner Sanctuary on their southward migration, and hence are not banded, or that the kill in the St. Marks area is relatively small. From about 10,000 geese wintering along a 100-mile stretch of coast during the late twenties, the annual kill was said to be several hundred (letter to Jack Miner from R. G. Porter, Apalachicola, Florida, winter of 1927-28). Future Status Although the Canada geese of the South- east population winter over an enormous area, extending from Virginia to Alabama and the Gulf Coast of Florida, their total number is not large. With the exception of the flock in the St. Marks area, most of the concentrations can be classified as being either small to medium in size and, in the aggregate, may match the St. Marks flock in size. Therein may lie the security of the Southeast population. The small concentra- tions, by virtue of their size, do not attract other than local hunters, whose kill is probably fairly light. The paucity of band recoveries from the St. Marks area suggests the possibility that the flock there is afforded adequate protection by the St. Marks Na- tional Wildlife Refuge. In any management measures involving the Southeast population, recognition should be given to the fact that the scattered flocks are but segments of a more or less con- tiguous population on the breeding grounds. These segments should be carefully censused at the time of the annual January inventory and the extent of the kill in each wintering area should also be determined within fairly close limits. To help insure the perpetua- tion of this population, it may be necessary to declare at least a portion of all reservoirs important to wintering geese, and some ad- jacent land areas, inviolate to hunting. Insofar as their habitat requirements in winter are concerned, Canada geese can be considered adaptive birds. They are quick to respond to changing agricultural prac- tices, to the creation of reservoirs, and to the formation of new refuges by changes in their habits and their local distribution. W. P. Baldwin, Jr., reported (personal communication, March, 1949) that in- creasing numbers of Canada geese are win- tering in northern Georgia, where they are resorting to the cultivated fields. At least some of these geese in former years must have migrated down to the St. Marks area. Such "reshuffles" in the population and the problems that arise from them should be recognized in any attempts to manage the Southeast geese. I APPENDIX B CLASSIFICL\TION OF THE CANADA GEESE OF THE GENUS BRANT A PROBABLY few other groups of North American birds have presented the taxonomists with greater challenge than the white-cheelced geese of the genus, Bratita. Before the distribution and the relationships of the various races can be fully understood, much collecting and banding will have to be dune on the breeding grounds. The com- plexity of the problem is apparent when it is realized that the race Braiita canadensis interior alone can be broken down into four fairly distinct breeding populations. As might be expected, the literature on the genus is fairly voluminous and often con- tradictory. Some plumage variations once thought to have taxonomic significance have been shown to be merely variations within single populations (Taverner 1931, Elder 1946. Hanson 19494). In the latest re- vision of the genus by Hellmayr & Conover (1948), the characters of the downy plum- age were taken into consideration. This factor considerably enhances the reliability of their study over studies previously made. They list the various members of the genus as follows : Branta leueopareia leiuopareia (Brandt). Tundra goose. [The lesser Canada goose of Kort- right (1942) and others.] Branta leueopareia occidenialis (Baird). West Coast goose. [The Western Canada goose of Kortright (1942) and others.] Branta minima Ridgway. Cackling goose. Branta canadensis parvipes (Cas- sin). Lesser Canada goose. [See Aldrich (1946) regarding the res- urrection of parvipes.^ Branta canadensis moffitti Aid- rich. Great Basin Canada goose. Branta canadensis interior Todd. Todd's Canada goose. Branta canadensis canadensis (Linnaeus). Eastern Canada goose. Branta hutchinsii (Richardson). Richardson's goose. [Branta cana- densis hiitchinsi of Kortright (1942) and others and sometimes known as Hutchins's goose.] Taverner (1931) has pointed out that several of the races are markedly distinct in the field, but as skins in the laboratory they are separated only with difficulty. According to James Mark, an Indian living at Eastmain, four different kinds of Canada geese are recognized by the James Bay Indians. The bird called Muskego nisku by the Cree Indian, meaning "large swamp goose," is the breeding goose of the muskeg, Branta canadensis interior, fig. 82. The "coast goose," ll'innipego nisku, is restricted to the James Bay coasts and observed only while on migration. It is reported as being smaller than the swamp goose, more vociferous, and having a rela- tively shorter neck, a description that fits the lesser Canada goose, Branta leueopareia leueopareia. Richardson's goose, Branta hutchinsii is called Apichishkish, meaning literally a small goose that has attained its full growth, fig. 82. The fourth kind recognized by the Indians on the south and east coasts of James Bay is described as being the largest of the group and possessing a brown breast, a feature from which it has derived its name, Kaoosoupasau-iit nisku. Geese of this kind are reported to breed farther north and are called the Fort George (Quebec) geese by the Moose Indians. The brown breast may represent staining by iron-rich waters of the areas frequented by this bird, which may possibly be B. c. interior. It is of interest to note that Blakiston (1863) also reported that an Indian on the Saskatchewan River described four dif- ferent kinds of "grey geese," the common gray goose, a short-necked goose, a small goose, and a large goose, descriptive names that roughly fit the forms described by the Indians of James Bay. The chief of the Indians around Lake St. Martin, Manitoba, told Taverner [203] Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Fig. 82.—Profile views of (upper head) an interior Canada goose, Branta canadensis interior and (lower head) a Richardson's goose, Branta Initchinsii. Both specimens are juvenile males. (Shortt & Waller 1937) that three kinds of Canada geese visit their area. The descrip- tions of these three varieties fit canadensis, leucopareia, and hutchinsii. According to Taverner's unpublished notes, which Shortt & Waller quote, an immense kind of Canada goose is also traditional with these Indians and "is so rare that it is known only by report. It is probably mythical." Despite Taverner's disbelief at one time in the reality of a very large goose, Mer- shon (1925) leaves little doubt that a very large variety of honker existed. McAtee (1944) has also commented on records of large geese from the Plains region. Aid- rich (1946) has now recognized this large race of Canada geese, giving it the name moffitii. Individuals of this race, presum- ably adult males, are known to range as high as 14 to 16 pounds, and even greater weights than these have been reported. Elder (1946) weighed 2,179 geese and the senior author weighed several thousand more geese at Horseshoe Lake, Illinois, without encountering an individual that at- tained the weight of 12 pounds. In the light of our present knowledge, the very large, almost legendary Canada goose known to many Indian groups in the boreal forest of Canada might be explained by individuals of the race Branta canadensis moffitti that have been occasionally taken north of their normal range. Such occasional invasions of the breeding grounds of one subspecies of Canada geese by nonbreeding members of another adjacent subspecies would not be unexpected. (In the above case the invasion of the range of B. c. in- terior by individuals of B. c. moffitti or an even larger extinct variety.) In the summer of 1949, Peter Scott, British ornithologist, and the senior author observed several flocks of nonbreeding "honkers," B. c. jnoffitti or interior, in the Perry River (Northwest Territories) breeding grounds of the smaller tundra Canada goose, skins of which have been identified by the senior author as those of Branta leucopareia leucopareia. LITERATURE CITED Adier. Frederick E. W. 1944. Chemical analyses of organs from leail-poisoneil Canada geese. Jour. Wildlife Mgt. 8(l):83-5. .\ldrich, John W. 1946. Speciation in the white-cheeked geese. Wilson Bui. 58(2) :94— 103. Allen, A. A. 1945. Some changes in the bird life of Churchill, Manitoba. ,\uk 62(1) :129-34. Anonymous 1901. Report of the survey and exploration of northern Ontario, 1900. Legislative Assembly of Ontario. 315 pp. .\skins, Major Charles 1945. What is good shooting? Sports .Afield 114(6) :48-56. .\udubon, John James 1843. The birds of America, vol. 6. 456 pp. Roe Lockwood i: Sons. New York. Austin, Oliver L. 1932. The birds of Newfoundland, Labrador. Memoirs Nuttall Ornith. Club 7:1-229. Cambridge, Mass. 1942. The life span of the common tern (Sli-rria hinnuin). Bird-Banding 13:159-76. Bailey, Alfred M., and Earl G. Wright 1931. Birds of southern Louisiana. Wilson Bui, 43(2 1 :114-42. Bailey, Harold H. 1913. The birds of Virginia. J. P. Bell Co., Lynchburg, Va. 362 pp. Baillie, James L., Jr., and Paul Harrington 1937. The distribution of breeding birds in Ontario. Roy. Can. Inst. Trans. 21 ( 1 ) :l-50. Barnston, George 1862. Recollections of the swans and geese of Hudson's Bay. Zoologist 20:7831-7. Baumgartner, Frederick M. 1942. An analysis of waterfowl hunting at Lake Carl Blackwell, Payne County, Oklahoma, for 1940. Jour. Wildlife Mgt. 6(1) :83-91. Bell, Robert 1880. Report on explorations on the Churchill and Nelson rivers and around God's and Island lakes, 1879. Append. \T. List of birds from the region between Norway House and Forts Churchill and York. Can. Geol. Surv. Rep. of Prog., 1878-79, pp. 67-70. 1883. Notes on the birds of Hudson's Bay. Roy. Soc. Canada Proc. and Trans., 1882 and 1883, 1(4) :49-54. 1887. Report on an exploration of portions of the -At-ta-wa-pish-kat &: Albany Rivers, Lonely Lake to James' Bay. Can. Geol. and Nat. Hist. Surv. Ann. Rep. 1886, part G. 39 pp., append. Bellrose. Frank C, Jr. 1947. .Analysis of methods used in determining game kill. Jour. Wildlife Mgt. 11(2) :105-19. Bellrose, Frank C, Jr., Harold C. Hanson, and Paul D. Beamer 1945. .Aspergillosis in wood ducks. Jour. Wildlife Mgt. 9(4):325-6. Bent, Arthur Cleveland 1925. Life histories of North American wihl fowl. U. S. Natl. Mus. Bui. 130. 311 pp. Berry, John 1939. The status and distribution of wild geese and wild ducks in Scotland. International Wildfowl Inquiry, vol. 2. Cambridge University Press. 190 pp. Bertram, G. C. L., David Lack, and B. B. Roberts 1934. Notes on east Greenland birds, with a discussion of the periodic non-breeding among Arctic birds. Ibis (series 13) 4(4), part 2, 816-31. [205] 306 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Bird, C. G., and E. G. Bird 1940. Some remarks on non-breeding in the Arctic, especially in north-east Greenland. Ibis (series 14) 4(4) :671-8. Blakiston, Captain Thomas 1863. On the birds of the interior of British North America. Ibis 5(7) :39-87. Brandt, Herbert 1943. Alaska bird trails. Bird Research Foundation, Cleveland, Ohio. 464 pp. Buss, Irven O. 1946. Wisconsin pheasant populations. Progress report of pheasant investigations con- ducted from 1936 to 1943. Federal aid in wildlife restoration project no. 9R. Wis. Cons. Dept., Madison, Wis. Pub. 326, A-46. 184 pp. Clarke, C. H. D. 1940. A biological investigation of the Thelon Game Sanctuary. Natl. Mus. Can. Bui. 96. 135 pp. Cooper, John M. 1933. Aboriginal land holding systems. Ms. memo to Dr. Harold W. McGill, Dept. Supt. General for Indian Affairs, Canada, Oct. 11, 1933. Darby, Perry C. 1916. Goose shooting on the Missouri River. }n Tales of duck and goose shooting, pp. 37-42. Chicago. Eastman Bros. Press for W. C. Hazelton. Dow, Jay S. 1943. A study of nesting Canada geese in Honey Lake Valley, California. Calif. Fish and Game 29(1) :3-18. Dutcher, William 1885. The Canada goose. Auk 2(1) :111. Elder, William H. 1946. Age and sex criteria and weights of Canada geese. Jour. Wildlife Mgt. 10(2) :93-lll. Ells, Sydney C. 1912. Report on James Bay surveys, exploration: Cochrane to James Bay, June 9 to September 12, 1911. Temiskaming and Northern Ontario Railway Commission. Legislative Assembly of Ontario, Toronto. 36 pp. Emlen, John T., Jr. 1940. Sex and age ratios in survival of the California quail. Jour. Wildlife Mgt. 4(1) :92-9. Errington, Paul L., and Logan J. Bennett 1933. Lost legions, a study of losses by crippling. Outd. Life 72(3) :18-9, 56. Farner, Donald S. ' 1945. Age groups and longevity in the American robin. Wilson Bui. 57(1) :56-74. Flower, Stanley S. 1925. Contributions to our knowledge of the duration of life in vertebrate animals.—IV. Birds. London Zool. Soc. Proc, pp. 1365-1422. Forbush, Edward Howe 1925. Birds of Massachusetts and other New England states. Part 1. Water birds, marsh birds and shore birds. Mass. Dept. Ag. 1:1-481. Gabrielson, Ira N. 1943. Wildlife refuges. Macmillan Company, New York. 257 pp. 1944. Managing the waterfowl. N. Am. Wildlife Conf. Trans. 9:264-9. Gillham, Charles E. 1948. Don't blame the Eskimos. Field and Stream 51 (12) :48-9, 151-4. Graham, Robert, and Frank Thorp, Jr. 1931. A laryngotracheitis syndrome in wild goose associated with pneumomycosis. Am. Vet. Med. Assn. Jour. 79(n.s. 32) (l):90-4. March, 1950 Hanson & Smith: Canada Geese 207 Grinnell, George Bird 1901. American duck shooting. Forest ami Stream Pub. Co., New York. 627 pp. Grinnell, Lawrence I., and Ralph S. Palmer 1941. Notes on bird-life of Churchill, Manitoba. Can. Field Nat. 55(4);47-5+. Hanson, Harold C. 1949a. Methods to determine age in Canada geese and other waterfowl. Jour. Wildlife Mgt. 13(2) :177-83. 1949A. Notes on white spotting and other plumage variations in geese. Auk 66(2) :164—71. 1949f. Trapping and handling Canada geese. III. Nat. Hist. Surv. Biol. Notes 20. 8 pp. Hawkins, Arthur S., and Frank C. Bellrose 1939. The duck flight and kill along the Illinois River during the fall of 1938. Am. Wild- life 28(41 :178-86. Hellmayr, Charles E., and Boardman Conoxer 1948. Catalogue of birds of the Americas and the adjacent islands: Part 1, no. 2 (Sphenis- cidae-Anatidae). Field Mus. Nat. Hist. Zool. Ser. 13. Pub. 615. 434 pp. Hess, Quimby 1943. Canada geese summering in northern Ontario. Can. Field Nat. 57(2-3) :46. Hewitt, C. Gordon 1921. The conservation of the wild life of Canada. C. Scribner's Sons, New York. 344 pp. Hochbaum, H. Albert 1944. The canvasback on a prairie marsh. American Wildlife Institute, Washington D. C. 201 pp. Hopkins, Ralph 1947. Quarterly progress report: Waterfowl management research. Wis. Wildlife Res. 5(4):12-'33. Horring. R. 1937. Report of the Fifth Thule expedition 1921-24. The Danish Expedition to Arctic North America, in charge of Knud Rasmussen, Ph.D. Gyldendalske Boghandel, Nordisk Forlag, Copenhagen, 2(6-9) :1-134. Howard, William J. 1934. Lead poisoning in Branta canadensis canadensis. Auk 51(4) :513^-. Howell, Arthur H. 1924. Birds of Alabama. Ala. Dept. Game and Fish. Brown Printing Co., Montgomery. 384 pp. Howley, James P. 1884. The Canada goose (Bernicia canadensis). Auk 1(4) :309-13. Huxley, Julian S. 1932. Field studies and physiology: a further correlation. Nature 133(3366) :688-9. Jackson, Hartley H. T. 1943. Conserving endangered wildlife species. Wis. Acad. Sci., Arts, Letters Trans. 35:61-89. Jenkins, Dale W. 1944. Territory as a result of despotism and social organization in geese. Auk 61(1) :30-47. Johnson, C. S, 1947. Canada goose management, Seney National Wildlife Refuge. Jour. Wildlife Mgt. ll(l):2I-4. Keith, David B. 1937. The red-throated diver in North East Land. Brit. Binis 31(3)66-81. Kendeigh, S. Charles 1942. .Analysis of losses in the nesting of birds. Jour. Wildlife Mgt. 6(1) :19-26. Kortright, Francis H. 1942. The ducks, geese and swans of North .'\merica. Am. Wildlife Inst., Washington, D. C. 476 pp. 208 Illinois Natural History Survey Bulletin Vol. 25, Art. 3 Lack David 1933. Nesting conditions as a factor controlling breeding time in birds. London Zool. Soc. Proc. 1-2, 231-7. 1941. Some aspects of instinctive behaviour and display in birds. Ibis fourteenth ser. 5{3):407-41. Leffingwell, William B. 1890. Wild fowl shooting. Rand McNally Co., Chicago. 373 pp. Leopold, Aldo 1931. Report on a game survey of the North Central States. Sporting Arms and Ammuni- tion Manufacturers' Institute, Madison, Wis. 299 pp. Leopold, Aldo, and Sara Elizabeth Jones 1947. A phenological record for Sauk and Dane counties, Wisconsin, 1935-1945. Ecol. Monog. 17:81-122. Leopold, Aldo, Theodore M. Sperry, William S. Feeney, and John A. Catenhusen 1943. Population turnover on a Wisconsin pheasant refuge. Jour. Wildlife Mgt. 7(4) :383- 94. Lincoln, Frederick C. 1934. The operation of homing instinct. Bird-Banding 5(4) :149-55. 1935. The waterfowl flyways of North America. Am. Game Conf. Trans. 21:264—76. 1939. The migration of American birds. Doubleday, Doran & Co., Inc., New York. 189 pp. Lloyd, Hoyes 1923. Observations on the wintering flocks of Canada geese in Nova Scotia. Can. Field Nat. 37(2) :26-8. Lorenz, Konrad Z. 1937. The companion in the birds' world. Auk 54(3) :245-73. Low, A. P. 1896. Report on the explorations in the Labrador Peninsula along East Main, Koksoak, Hamilton, Manicuagan and portions of other rivers in 1892-93-94—95. Can. Geol. Surv. Ann. Rep., new series, vol. 8, rep. L, append. II (List of birds of the interior of the Labrador Peninsula), pp. 323-8. 1902. Report on an exploration of the east coast of Hudson Bay from Cape Wolstenholme to the south end of James Bay. Geol. Surv. Canada Ann. Rep., new series, vol. 8, rep. D., no. 778, pp. 1-84. Low, Seth H. 1935. A coastal group of Canada geese. Bird Banding 6(2) : 67-8. Lower, A. R. M. 1915. A report on the fish and fisheries of the west coast of James Bay. In sessional paper 39rt. Forty-seventh annual report of the Department of Marine & Fisheries 1913-14. Fisheries 50(27) :31-67. Manniche, A. L. V. 1910. The terrestrial mammals and birds of north-east Greenland. Meddelelser om Grjin- land 45:1-199. Manning, T. H. 1946. Bird and mammal notes from the east side of Hudson Bay. Can. Field Nat. 60(4)71-85. McAtee, W. L. 1924. Do bird families have any permanency? Condor 26(5):193-4. 1944. Popular subspecies. Auk 61 (1) :135-6. McClanahan, Robert C. 1940. Original and present breeding ranges of certain game birds in the United States. U. S. Biol. Surv. Wildlife Leaflet BS-158. 21 pp. Mcllhenny, Edward A. 1940. An early experiment in the homing ability of wildfowl. Bird-Banding 11(2) :58-60. 1 AJiirch, 1950 Hanson & Smith: Canada Geese 209 Mershon, William B. 1925. Big geese. Field and Stream 34:26-7, 63-4. Miner, Jack 1923. Jack Miner and the birds. Reilly & Lee Co., Chicago. 176 pp. 1929. Jack Miner on current topics. Ryerson Press, Toronto, Can. 111pp. Miner, Manly F. 1931. Migration of Canada geese from the Jack Miner Sanctuary and banding operations. Wilson Bui. 43(1) :29-34. Moffitt, James 1935. Waterfowl shooting losses indicated by banding returns. Am. Game Conf. Trans. 21 :305-8. Montgomery, George A. 1938. More wings for tomorrow. Capper's Farm. 49(10) :12-3. .\Iunro, J. A. 1936. A study of the ring-billed gull in .'\lberta. Wilson Bui. 48(3) :169-80. Nelson, Edward W. 1887. Report upon natural history collections made in Alaska between the years 1877 and 1881. III. Arctic series of publications issued in connection with the Signal Service, U. S. Army, 49th Congress, 1st Sess. Misc. Doc. 156, no. 3. Government Printing Office, Washington, D. C. 337 pp. Pearson, Gilbert T., C. S. Brimley, and Herbert Hutchison Brimley 1942. Birds of North Carolina. N. C. Dept. .'Vg. State Museum. 416 pp. Phillips, John C. 1916. Two problems in the migration of water fowl. I. Do American ducks reach the Marshall Islands? II. Behavior and makeup of the migrating flocks of Canada geese. Auk 33( 1) :22-7. 1921. Massachusetts geese. Auk 38(2) :271-2. Phillips, John C., and Frederick C. Lincoln 1930. American waterfowl. Their present situation and the outlook for their future. Houghton Mitflin Company, Boston and New York, 327 pp. Pickens, A. L. 1928. Birds of upper South Carolina: A study in geographical distribution. Wilson Bui. 40(3) :182-91 (35 n.s., whole no. 144). Pirnie, Miles D. 1935. Michigan waterfowl management. Mich. Dept. Cons., Lansing. 328 pp. 1939. Wildlife conservation. Wilson Bui. 51 (4) :246-7. Preble. Edward A. 1902. A biological investigation of the Hudson Bay region. North American Fauna 22:1-140. U. S. Dept. Ag. Div. Biol. Surv. Government Printing Office, Washington, D. C. Renison, Robert John 1944. The miracle of spring. In Wednesday morning. McClelland & Stewart, Toronto. Richardson, Sir John 1851. Arctic searching expedition, vol. 2. 426 pp. Longmans, Brown, Green, and Long- mans, London. Rousseau, Jacques 1948. By canoe across the Ungava Peninsula via the Kogaluk and Payne rivers. Arctic 1(2) :133-5. Shortt, T. M., and H. S. Peters 1942. Some recent bird records from Canada's eastern Arctic. Can. Jour. Res. 20(D) :338-48. Shortt, T. M., and Sam Waller 1937. The birds of the Lake St. Martin region, Manitoba. Roy. Ont. Mus. Zool. Contrib. 10:1-51. 210 Illinois Natural History Survey Bulletin Vol. 25, Art. .3 Soper, J. Dewey 1930. The blue goose; an account of its breeding ground, migration, eggs, nests and general habits. Can. Dept. Interior, Ottawa. 64 pp. 1946. Ornithological results of the Baffin Island expeditions of 1928-1929 and 1930-1931, together with more recent records. Auk 63(1) :l-24. Stone, Witmer 1937. Bird studies at old Cape May. An ornithology of coastal New Jersey. Del. Valley Ornith. Club at Acad. Nat. Sci. Phila. 520 pp. Sutton, George M. 1932. The exploration of Southampton Island, Hudson Bay. Part II, Zoology, section 2: The birds of Southampton Island. Carnegie Mus. Mem. 12(2) :3-268. Taverner, P. A. 1922. Adventures with the Canada goose. Can. Field Nat. 36(5) :81-3. 1931. A study of Branla canadensis (Linnaeus), the Canada goose. Natl. Mus. Can. Bui. 67:28-40. Taverner, Percy A., and George Miksch Sutton 1934. The birds of Churchill, Manitoba. Carnegie Mus. Ann. 23(1) :l-83. 14 pis. Tinbergen, N. 1942. An objectivistic study of the innate behavior of animals. Bibliotheca Biotheoretica, Series D, l(2):39-98. 1948. Social releasers and the experimental method required for their study. Wilson Bui. 60(1):6-S1. Todd, W. E. Clyde 1938. A new eastern race of the Canada goose. Auk 55(4) :661-2. 1940. Birds of western Pennsylvania. University of Pittsburgh Press. 710 pp. Tufts, R. W. 1932. Annual convention of winter geese. Can. Field Nat. 46(3) :51-3. Uhler, Francis M. 1933. Effect of baiting and live decoys on the waterfowl of the upper Mississippi River valley. U. S. Fish and Wildlife Service. (Unpublished report.) Wetmore, Alexander, and Committee 1945. Twentieth supplement to the American Ornithologists' Union check-list of North American birds. Auk 62(3) :43 6^9. Wilfrid, Rev. Brother 1924. Notes on the Canada goose in captivity. Can. Field Nat. 38(7) :124. Williams, Cecil S., and E. R. Kalmbach 1943. Migration and fate of transported juvenile waterfowl. Jour. Wildlife Mgt. 7(2):163-9. Williams, Cecil S., and William H. Marshall 1938. Survival of Canada goose goslings. Bear River refuge, Utah, 1937. Jour. Wildlife Mgt. 2(l):17-9. Williams, Cecil S., and Marcus C. Nelson 1943. Canada goose nests and eggs. Auk 60(3) :341-5. Whitlock, S. C, and H. J. Miller 1947. Gunshot wounds in ducks. Jour. Wildlife Mgt. 11 (3) :279-81. Yeager, Lee E., and William H. Elder 1945. Pre- and post-hunting season foods of raccoons on an Illinois goose refuge. Jour. Wildlife Mgt. 9(l):48-56. Zimmerman, P. R. 1942. Quarterly progress report: Waterfowl management research. Wis. Wildlife Res. 2(1) :17-31. 1943. Quarterly progress report: Waterfowl management research. Wis. Wildlife Res. 3(1) :15-21. Recent Publications A.—ILLINOIS NATURAL HISTORY SURVEY BULLETIN. Volume 22, Article 1.—^The Plant Bugs, or Miridae, of Illinois. By Harry H. Knigl September, 1941. 234 pp., frontis. + 181 figs., bibliog., index. $125. Volume 22, Article 2.—Studies of North American Plecoptera, with special reference the fauna of Illinois. By T. H. Prison. September, 1942. 122 pp., frontis. + 126 fig bibliog., index. $1.00. Volume 22, Article 6.—Survey of the Illinois Fur Resource. By Louis G. Brown ai Lee E. Yeager. September, 1943. 70 pp., frontis. + 33 figs., bibliog. (Bound wi Article 7.) Volume 22, Article 7.—Illinois Furbearer Distribution and Income. By Carl O^ Moh September, 1943. 33 pp., frontis. + 24 figs., bibliog. (Bound with Article 6.) Volume 23, Article 1.—^The Caddis Flies, or Trichoptera, of Illinois. By Herbert I Ross. August, 1944. 326 pp., frontis. + 961 figs., bibliog., index. $1.50. Volume 23, Article 2.—Duck Populations and Kill. By Frank C. Bellrose, Jr. Novell ber, 1944. 46 pp., frontis. + 27 figs., bibliog. 50 cents. Volume 23, Article 3.—Overfishing in a Small Artificial Lake : Onized Lake near Alto Illinois. By George W. Bennett. May, 1945. 34 pp., frontis. + 15 figs., bibliog. Volume 23, Article 4.—^Wetwood of Elms. By J. Cedric Carter. August, 1945. 42 p| frontis. + 30 figs., bibliog. Volume 23, Article 5.—Fox Squirrels and Gray Squirrels in Illinois. By Louis ( Brown and Lee E. Yeager. September, 1945. 88 pp., frontis. + 42 figs., bibliog. Volume 24, Article 1.—The Mosquitoes of Illinois (Diptera, Culicidae). By Herbe H. Ross. August, 1947. 96 pp., frontis. + 184 figs., bibliog. 50 cents. Volume 24, Article 2.—^The Leafhoppers, or Cicadellidae, of Illinois (Eurymelinai Balcluthinae). By D. M. DeLong. June, 1948. 280 pp. + 514 figs., bibliog., inde $1.25. Volume 24, Article 3. — ^The Bass-Bluegill Combination in a Small Artificial Lake. I George W. Bennett. December, 1948. 36 pp., frontis. + 10 figs. Volume 24, Article 4.—^The Pseudoscorpions of Illinois. By C. Clayton Ho£E. Jun 1949. 86 pp., frontis. + 51 figs., bibliog., index. 50 cents. Volume 25, Article 1.—Characteristics of Residual Insecticides Toxic to the House Fl By Willis N. Bruce. July, 1949. 32 pp., frontis. + 14 figs., bibliog. Volume 25, Article 2.—Eflect of Permanent Flooding in a River-Bottom Timber Are By Lee E. Yeager. August, 1949. 32 pp., frontis. +21 figs., bibliog. B.—ILLINOIS NATURAL HISTORY SURVEY CIRCULAR. 32.—Pleasure Vi^ith Plants. By L. R. Tehon. February, 1949. (Third printing, vrh revisions.) 32 pp., frontis. + 9 figs. 36.—Planting and Care of Shade Trees. By J. E. Davis. September, 1947. (Thii printing, with additions.) 28 pp., frontis. + 20 figs. 38.—^Windbreaks for Illinois Farmsteads. By J. E. Davis. February, 1949. (Thii printing, with additions by L. B. Culver.) 33 pp., frontis. + 27 figs. 39.—How to Collect and Preserve Insects. By H. H. Ross. July, 1949. 59 pi frontis., + 65 figs. 41.—How to Recognize and Control Termites in Illinois. By B. G. Berger. Februat; 1947. 44 pp., frontis. + 32 figs. 42.—Bird Dogs in Sport and Conservation. By Ralph £. Yeatter. December, 194 64 pp., frontis. + 40 figs. C—ILLINOIS NATURAL HISTORY SURVEY MANUAL. 2.—Fieldbook of Illinois Land Snails. By Frank Collins Baker. August, 1939. 166 pj color frontis. + 170 figs., 8 pis. $1.00. 3.—Fieldbook of Native Illinois Shrubs. By Leo R. Tehon. December, 1942. 307 pf 4 color pis. + 72 figs., glossary, index. $1.25. List of avmlabU publications, about 400 titlts, mailed on request. Single copies of Illinois Natural History Survey publications for which no price is liitc will be furnished free of charge to individuals until the supply becomes low, after which nominal charge may be made. More than one copy of any free publication may be obtaioi without cost by educational institutions and official organizations within the State of Illinoil prices to others on quantity orders of these publications will be quoted upon request. Address orders and correspondence to the Chief Illinois Natural History Survey Natural Resources Building, Urbana, Illinois Payment in the form of U.S. Post Office money order made out to State Treasurer of Illiaoii Springfield, Illinois, must accompany requests for those publications on which a price ii set.